Home

TESTING THERMAL IMAGERS - Practical Guide

1. Radio Microwave Infrared Visible Ultraviolet X Ray LONG wavelength eee tees cece ee ce eee eee nnee mrnte mennenm nnmnnn m ee ee Short wavelength Low frequency e eecececeeeeaeaesnencecerereeae se snenenserereoaesesnenenseeeeeoas High frequency Fig 1 1 Spectrum of electro magnetic radiation Electro optical imaging systems are the systems that 1 create visible image of the targets being observed 2 use optical radiation emitted or reflected by the targets to get information necessary to create the output image 3 convert coming optical radiation into electrical signal process signal con vert electrical signal into visible image or into electrical equivalent of vis ible image Human eye can be treated as a type of electro optical imaging system because it uses visible radiation reflected by the targets converts incoming optical radiation into electrical signals processes signal in a brain and finally converts electrical sig nals into electrical equivalent of a visible image understood by a human brain 1 2 Human sight Sensing of visible radiation light by a human eye provides about 90 inform ation coming to the human brain which makes sight the most important of all senses In spite of its importance the phenomenon of human sight is complex and still not fully explained However to simplify it we can assume that in general the set the biological eye optic nerves an
2. AT step AT step below 0 5 C 0 01 C 1 C 2 0 C 0 1 C 0 5 C 1 0 C 0 02 C 2 C 8 0 C 0 2 C 1 C 2 0 C 0 05 C over 8 0 C 0 4 C 126 Table 5 5 Updated recommended steps of differential temperature during MRTD measurement AT step AT step lt 0 02 C 0 002 C 0 5 C 1 0 C 0 02 C 0 02 C 0 1 C 0 005 C 1 C 2 0 C 0 05 C 0 1 C 0 5 C 0 01 C 1 C 2 0 C 0 1 C 5 1 9 Measurement methods There are two basic methods to measure MRTD 1 Keep constant spatial frequency and vary temperature difference 2 Keep constant temperature difference and vary spatial frequency In case of classical test systems variable target test system only the first method is possible In case of variable distance test systems both methods can be used but practically the second method is applied In case of the first method in order to measure MRTD at one measurement point spatial frequency is kept constant but temperature difference varies Practically this means that a single target is used but the observer regulates temperature difference between the bars and the background until he finds that he can recognize the 4 bar pattern The procedure is repeated by several observers and then a target is ex changed and measurement of anew MRTD point is carried out In case of the second method in order to measure MRTD at one measurement point tempe
3. 131 Table 5 7 Exemplary table with horizontal MRTD test results Observer I Observer II Observer III Nia v AT AT AT AT AT AT AT AT AT MRTD MRTDo o 112 00 0 50 0 02 0 08 0 05 0 02 0 08 0 05 0 03 0 09 0 06 0 05 0 05 2 1 00 1 00 0 05 0 11 0 08 0 07 0 13 0 1 0 09 0 15 0 12 0 1 0 09 3 0 67 1 50 0 13 0 19 0 16 0 17 0 23 0 2 0 22 0 28 0 25 0 2 0 19 4 0 50 2 00 0 27 0 33 0 3 0 34 0 4 0 37 0 37 0 43 0 4 0 36 0 33 5 0 40 2 50 0 47 0 53 0 5 0 55 0 61 0 58 0 61 0 67 0 64 0 57 0 53 6 0 33 3 00 0 77 0 83 0 8 0 84 0 9 0 87 0 87 0 93 0 9 0 86 0 79 7 0 29 3 51 1 47 1 53 1 5 1 47 1 53 1 5 1 77 1 83 1 8 1 6 1 47 8 0 25 4 00 3 17 3 23 3 2 2 97 3 03 3 3 47 3 53 3 5 3 23 2 98 a bar width in mm v spatial frequency in lp mrad MRTD in C units AT temperature difference in C units Table 5 8 Exemplary table with vertical MRTD test results Observer I Observer I Observer III Nia v AT AT AT AT AT AD AT AT AT MRTDn n MRTD or o 1 2 00 0 50 0 03 0 09 0 06 0 02 0 08 0 05 0 03 0 09 0 06 0 06 0 05 2 1 00 1 00 0 06 0 12 0 09 0 05 0 11 0 08
4. Fig 4 25 Concept of calibration of measurement thermal imagers 4 3 3 Boresighting to a reference optical mechanical axis Boresighting is a process to align optical axis of a single system or a series ofoptical or electro optical systems with a certain reference optical axis or mechanical axis Basically the aim of boresighting of a thermal imager is to achieve a situation when a optical axis of the thermal imager is parallel to axis of other electro optical surveillance systems the case when the thermal imager is a part of a bigger multi sensor surveillance system b optical axis of the thermal imager is parallel to a reference mechanical axis In other words the aim of boresighting process is to achieve a situation when a all subsystems of bigger electro optical imaging systems are looking into the same point assumption the distance is hundreds times bigger than the focal length of the optics of the subsystems b an imaging system the thermal imager is looking at a point of exactly known coordinates relative to the reference mechanical axis When the boresighting process is not carried out properly we can get a situation when 113 a the thermal imager generates an image of a slightly different area that the area seen on the image produced by the TV camera the two images are slightly displaced laser range finder can measure a distance but not to the point marked on the image generated by the thermal imag
5. ccsscssssscsecssscscseescsssssecscsssssssscsesssssssesssseees 147 5 4 Noise PAFAMEUELS ise csiscaissiesscsdenscsteossvssadvedsdsssberessadcesstensdsiesesssdesasecctecdsessobseseveeuses 150 SAA Measurement concept rines veatessavevsventavaneveseseesadenstiantvnccedan saeshane E 150 5 4 2 Measurement procedure cccccccesesceeseessceseeseeeseeeeeeseceeeeaeceeeeseeeseseneeensneeeses 152 5 4 3 Interpretation of measurement results eeeeseeeeseeeseeseeneeeseeseeeeeeeeneeeees 155 5 5 Referentes vcccsssssccessssiassseesssaseseasecnaansascdesesseesadeconaessusedoasteansesassaescancesustsecnevese scnss 156 6 STANAG 4349 requirements cccccccceeseeeeeeeeeeeeeeeeeeeeeeeeees 157 7 Guidelines for buyers of test equipment c 162 vi 1 Introduction 1 1 Concept of electro optical imaging system According to the International Lighting Vocabulary published by the Internation al Lighting Commission CIE and the International Electrotechnical Commission CIE 15 considered nowadays as an international primary authority on terminology in radiometry electromagnetic radiation between radio radiation and X radiation is termed the optical radiation Thus the optical radiation can be defined as radiation of wavelengths longer than about 1 nm and shorter than about 1 mm The range of optical radiation is divided into three sub ranges infrared radiation visible radi ation and ultraviolet radiation
6. The variable target distance test systems Fig 4 6 Fig 4 7 are practically the variable target test systems without an IR collimator to project a thermal image gen erated by a set target blackbody Because there is no collimator and the generated thermal image must be seen directly by the tested thermal imager then the distance tested imager test system must be bigger than minimal focus distance of the tested imager At the same time typical commercially available rotary wheels are adapted to use target of small size typical diameter about 50 mm and the distance target imager must be short if typical imager is to see small details of the target This creates situation when only thermal imagers of short minimal focus distance can be tested using the variable target distance test systems Commercial thermal imagers are designed for observation of short distance targets minimal focus distance is typically about 0 5m and such imagers can always be tested using the discussed group of test systems However most surveillance military type imagers design for observation of long distance targets cannot not be tested using the variable tar get distance test systems due to much longer minimal focus distance typically over 5m When the distance imager target is over the minimal focus distance of the im ager then we cannot carry out parameters of measurement procedures that require simulation of big targets 76 Rotary wheel distance
7. 5 16 Method two an image of an edge target generated by tested imager is acquired analyzed and later MTF of tested imager is calculated as a module from Fourier transform of the differentiated Edge Spread Function normalized blur distribution in the image of the edge target MTF v F ESF x 5 17 Both methods are characterized by some advantages and disadvantages A Slit method 1 Low sensitivity to noise inherent in image to be analysed 2 Measurement result must be corrected depending on width of the slit used for measurement 3 Several slit targets are needed when testing imagers of different field of view 4 High temperature difference over 10 K or more is required as assure a clear image of the slit target 143 B Edge method 1 High sensitivity to sensitivity to noise inherent in image to be analysed 2 One edge target enables measurement of MTF of any type of thermal im agers 3 Low temperature difference over 3 K is required as assure a clear image of the edge target 4 It is technologically easier to manufacture high quality edge target than slit target Due to lower sensitivity to noise the slit target method was the preferred MTF measurement method in the past Noways manufacturers of equipment for testing thermal imagers generally prefer measurement of MTF of thermal imagers using edge target method 8 10 Therefore only the edge method shall be discussed l
8. Fig 4 7 Photo of the SAFT variable target distance measuring system All three types of earlier presented types of test systems posses some advantages and disadvantages The variable target test systems or variable target projectors of configuration shown in Fig 4 2 can be considered as classical systems for testing thermal im agers Due to high thermal inertia use of baffles in the collimator motorized rotary wheel and the fact that they are mostly used in laboratory conditions they are char acterized by very good stability and measurement accuracy Any thermal imager is able to focus on optical infinity distance simulated by the collimator Therefore all types of thermal imagers surveillance or commercial short distance of long distance can be tested using these test systems on condition that the aperture of the imager optics is smaller then output aperture of the collimat or Big size of these systems mostly due to the use a collimator is main draw back of the variable target test systems Even in case of smaller collimators it is dif ficult to use such a test system outside laboratory due transport problems and neces sity to align tested imager with the collimator output 77 The variable distance test systems of configuration shown in Fig 4 4 repres ent a class of test systems of much more compact design than in case of variable tar get systems Collimators and rotary wheels are eliminated Because of small size and
9. ager and has to be determined in each case 5 1 The observer shall have normal visu Requirement on users of test al acuity post correction defects less than equipment This requirement can 0 25 diopters and good color vision be fulfilled if proper observers are and be experienced in this type of meas chosen and trained urement 5 2 The test pattern shall be a plate con Targets for MRTD measurement taining four rectangular slots forming a are manufactured as a plate con square of four bars and three spaces taining four rectangular slots Length width ratio of the rectangular slot forming a square of four bars and is 7 1 three spaces Length width ratio of The space width should be equal to width the rectangular slot is 7 1 Target is of the rectangle The test pattern is posi fixed to a rotary wheel that is po tioned in front of a black body the tem sitioned in front of the blackbody perature of which can be varied giving a temperature difference at between the bars and the spaces 5 2 1 The spatial frequency of the Point 5 2 1 presents require resolving power measurement targets ments on manufacturing must be within 5 of the nominal tolerances of 4 bar targets and value tolerances of focal length of the 157 5 2 2 The emissivities of the test pattern and the black body must both be at least 0 95 5 2 3 The blackbody must make it pos sible to achieve temperature differences
10. server MRTD MRTD MRTD3 MRTD n 5 11 where MRTD is the measurement result for n observer Fill the MRTD column Correct the raw measurement MRTD results using this formula MRTDeo MRTDn cor T Teo 5 12 where cor T is the correction coefficient see Table 5 1 and to is trans mittance of the collimator Fill the MRTD column Table 5 9 Exemplary table with horizontal MRTD test results using the simplified procedure Observer I Observer II Observer III No ja v AT ATi AT MRTD MRTD cor 1 2 00 0 50 0 05 0 05 0 06 0 05 0 05 2 1 00 1 00 0 08 0 1 0 12 0 1 0 09 31 0 67 1 50 0 16 0 2 0 25 0 2 0 19 4 0 50 2 00 0 3 0 37 0 4 0 36 0 33 5 0 40 2 50 0 5 0 58 0 64 0 57 0 53 6 0 33 3 00 0 8 0 87 0 9 0 86 0 79 7 0 29 3 51 1 5 1 5 1 8 1 6 1 47 8 0 25 4 00 3 2 3 3 5 3 23 2 98 a bar width in mm v spatial frequency in lp mrad AT temperature difference in C units MRITD in C units 135 Table 5 10 Exemplary table with vertical MRTD test results using the simplified procedure Observer Observer II Observer III I No Ja v AT AT AT MRTDy MRTD cor 1 2 00 0 50 0 06 0 05 0 06 0 06 0 05 2 1 00 1 00 0 09 0 08 0 12 0 1 0 09 3 0 67 1 50 0 18 0 22 0 3 0 23 0 21 4 0 50 2 00 0 35 0 4 0 55 0 43 0 4 5 0 40 2 50 0 65 0 7 0 8 0 72 0 66 6 0 33 3 00 1 1 1 4 1 4 1 3 1 2 7 0 29 3 51 3 50 3 2 3 4 3 37 3
11. 13 Holst G C Testing and evaluation of infrared imaging systems JCD Publishing Company 1993 14 Holst G C The Infrared amp Electro Optical Systems Handbook Vol 3 Electro Optical System Design Analysis and Testing Chapt 4 pp 206 207 SPIE 1993 15 ISO 15529 Principles of measurement of modulation transfer function MTF of sampled imaging systems 1999 16 Jaccobs E Dynamic MRTD simulation Infrared Imaging Systems Design Analysis Model ing and Testing XI SPIE Vol 4030 pp 70 76 2000 17 Lloyd J The Infrared amp Electro Optical Systems Handbook Vol 3 Electro Optical System Design Analysis and Testing Chapt 1 SPIE 1993 18 MIL I 24698 SH Infrared thermal imaging systems Department of Defense USA 1988 19 MIL STD 1859 Thermal Imaging Devices Performance Parameters Of 1983 20 MIL T 49381 Test Set Thermal sight TS 3681 VSG USAERADCOM 1980 21 Miller Scott J Backer Brian S Kohin Margaret Alonso Pascual Whitwam Jason T Test methods and technology for uncooled imaging systems Infrared Imaging Systems Design Analysis Modeling and Testing XV SPIE Vol 5407 30 37 2004 22 Night Vision Thermal Imaging Systems Performance Model User s Manual amp Reference Guide U S Army Night Vision and Electronic Sensors Directorate AMSEL RD NV MS SPMD 2001 23 Pieper R Cooper A Celik M Kenter Y Objective MRTD experimental measurements compared with predictions based on the visibil
12. Bryant P Grigor J McHugh S Performance comparison of reflective and emissive target projector systems for high performance IR sensors Infrared Imaging Systems Design Analysis Modeling and Testing XIV SPIE Vol 5076 pp 242 249 2003 4 Chrzanowski K J Fischer W Wrona Testing of Thermal Imagers For Non Destructive Thermal Testing Applications ASTM Journal of Testing and Evaluation 28 395 402 2000 5 Chrzanowski K Evaluation of IR collimators for testing of thermal imaging systems Optoelectronics Review 1 2007 6 Chrzanowski K Lee H C Wrona W A condition on spatial resolution of IR collimators for testing of thermal imaging systems Optical Engineering 39 5 14137 1417 2000 7 Chrzanowski K Evaluation of commercial thermal cameras in quality systems Optical Engineering Vol 41 No 10 2002 8 Digital Signal Processing Application with the TMS320 Family Theory algorithms and implementations Vol 2 Texas Instruments 1990 570 573 9 Guide to the expression of uncertainty in measurement International Organisation for Standarisation International Electrotechnical Commission International Organisation of Legal Metrology International Bureau of Weights and Measures TAG 4 WG 3 1993 10 Holst G C Testing and evaluation of infrared imaging systems JCD Publishing Company 2008 114 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 3
13. Sur mxn each pixel is averaged 2D spatial noise over N frames 3 Sur nxN each column is aver temporal column noise rain aged over m pixels variations of mean column brightness with time 4 Spr mxN each row is averaged temporal row noise streak over n pixels ing variations of mean row brightness_with time 5 Sy m each row is averaged spatial row noise variations over n pixels and N of mean row brightness that frames do not depend on time 6 Sz n each column is aver spatial column noise vari aged over m pixels and ations of mean column N frames brightness that do not depend on time 7 Sr N each frame is averaged frame to frame brightness over mxn pixels variation flicker 8 S 1 each frame is averaged average brightness of the over mxn pixels and N frames in the sequence frames Table 3 6 Conversion formulas 3D TVH array to 2D VH array 3D TVH array to 2D TH array array N L Srl j k Syy i j is N i can vary from 1 to n ho rizontal E Sryy i j k Syr J k i j can vary from 1 to m vertical E Sryy Gs j k Spr Uk z 3D TVH array to 2D TV temporal 3D TVH to 2D TV array 2D TV array to 1D _V array 2D TH array to 1D T k can vary from 1 to N N m N n m n L Sra i j k YL Srah j k L E Sru li j k S j 4 y ed a a S a S k 3D TVH to a single number 46 3 3 2 Single paramete
14. called also night vision devices I2S improve hu man eye sensitivity and offer observation in most night conditions TV cameras generate high quality color or monochromatic images of the observed scenery in day conditions and ensure later recording processing and later transfer of these im ages using modern telecommunication equipment Some types of TV cameras of ten called LLTV cameras offer improved sensitivity and can be used in night con ditions for similar applications like NVDs Finally thermal cameras due to differ ent spectral ranges can work even at total darkness in bad visibility conditions and make camouflage much more difficult All types of the mentioned earlier imaging systems when working at narrow field of view provide visualization of small de tails of the scenery being observed and in this way improve spatial resolution of a human eye Electro optical imaging systems Night vision devices TV cameras Thermal imagers Fig 1 5 Basic types of electro optical imaging systems Night vision devices NVD are the imaging systems built using an image in tensifier tube consisting of a photocathode an anode in form of a phosphor screen and other optional components The tube intensifies a low luminance image of the observed objects created on the photocathode into a brighter image created on the anode NVDs are direct viewing devices that generate the output image by optical amplification of the input im
15. pixels in FPA sensor mean that thermal cameras built using this sensor should theoretically resolve smaller details However practically it is not always true The same is with DAS or IFOV parameters Lower values of DAS IFOV are welcome but they do not always indicate improvements in thermal camera ability to resolve small details 3 4 2 MTF and derivative parameters MTF modulation transfer function is a function of the contrast of image ofa sine pattern at a given spatial frequency generated by the tested camera relative to a contrast of an image of sine pattern at spatial frequency equal to zero Spatial frequency is typically measured per cycles or line pairs per a unit angle or a unit length in case of thermal cameras MTF in line pairs per millionaires LP mrad or in inverted millionaires mrad see Fig 3 25 sine pattem l a pattem oyde Tx e uar lt UZ ON spatial frequency distance R v 1 6 Fig 3 25 Graphical interpretation of spatial frequency Images of several sine patterns of different spatial frequency generated by a thermal camera of MTF presented in Fig 3 26 are shown in Table 3 9 Attention TV resolution is measured in line widths instead of pairs where there are two line widths per pair over the total height of the display 56 MTF 0 9 0 8 0 7 0 6 0 4 0 3 0 1 4 6 spatial frequency Ip mrad 8 10 Fig 3 26 Exemp
16. space imagers is a true technical art where many small details like type of mirror mount are of utmost importance and there are special requirements on room where the collimator is to be used In case of collimators for testing thermal imagers the situation is much easier but still it is always recommended for users of collimat ors to calculate required spatial resolution of the collimator using formula 4 2 and to verify this condition by practical tests to be sure that the collimator projects the images of proper quality during tests of thermal imagers Table 4 4 Requirements on spatial resolution Veo of IR collimators to be used in testing different thermal imagers imager HFOVxVFOV horizontal V 0 lp mrad Angenieux 140mrad x105mrad sion 2000 436x314 mrad 105x75 3mrad NFOV 0 98 x0 71 9 36 46 8 Matiz long WFOV 6 53 x4 640x480 2 8 14 range 114x69 8mrad micro i id SAGEM scanning 23 7x15 9mrad Ultra 275C WFOV 18 x13 Gace ee ae FLIR Inc 314x227mrad NFOV 4 x2 89 69 8x50 4mrad TV camera FOV 1 6 4000 pixel 70 0 for space linear program CCD detector HFOV horizontal Field Of View VFOV vertical Field Of View 87 B Aperture Focal length Collimator aperture is the diameter of the maximal ray beam that can be gener ated by the collimator when a point source is used It is strictly needed for proper testing that a collimator aperture must be bigger than a diameter of the optics of
17. the standards and special ized literature recommend to carry out the measurements for two cases vertical bars so called horizontal MRTD and horizontal bars so called vertical MRTD and to average results In case of staring thermal imagers the difference between ho rizontal MRTD and vertical MRTD is usually small Totally inverse situation is in case of scanning thermal imagers Horizontal MRTD is usually significantly better 50 or more than vertical MRTD Fig 5 4 Therefore it is necessary to be very careful with a target position when testing scanning thermal imagers A simple 90 rotation is enough to change drastically test results Fig 5 4 Image of double 4 bar target generated by a scanning thermal imager 5 1 5 Offset International standards recommend to carry out MRTD measurements first for positive patterns later for negative pattern and finally to average results 1 4 The aim of this recommendation is to correct influence of offset effect on MRTD measurement results Offset exists when minimal positive temperature difference set by the blackbody controller needed to recognize positive 4 bar pattern of low frequency target is not equal to minimal negative temperature difference needed to recognize negative 4 bar pattern of the same target We can also less accurately say that offset exists when the observer can recognize 4 bar pattern in situation when the blackbody controller indicates differential temperature equal
18. 0 04 1 5 lt 0 35 C lt 0 1 C lt 0 25 C 0 16 2 5 lt 0 95 C lt 0 3 C lt 0 55 C 0 5 3 5 lt 3 5 C lt 1 3 C lt 1 9 C 1 7 Before we decide whether the imager fulfills MRTD requirements we must consider the fact that only measurement result with error uncertainty indicator is truly valid measurement data It is frequently reported in literature that during MRTD measurement variability as high 50 are often cited from laboratory to laboratory and 20 at one laboratory 3 Let us assume this worst case scenario of 50 measurement error MRTD measurement results with 50 measurement er rors of the analysed imager and exemplary three versions of MRTD requirements are shown in Fig 5 11 10 MRTD 0 1 0 01 0 5 1 1 5 2 2 5 3 frequency lp mrad 3 5 Fig 5 11 MRTD test results continuous curve with errors bars and three different versions of MRTD requirements version A triangles down version B triangles up version C squares version D double triangles It is clear that in case A the tested thermal imager fulfills the MRTD requirements It is also clear that in case C the tested imager does not fulfill the requirements There is however a question what should be a proper decision in case B and case D There is no clear answer what to do in this situation in 142 international standards and literature Totally different decisions can be taken on the basis of the presente
19. 1 8 0 25 4 00 The test procedure can be shorten even more if measurements are made by only one but well trained observer It is a risky solution but if it was earlier verified that the observer indications are close to group of average indications then the solution can be considered as acceptable B Case of variable distance test system Variable distance test systems are systems built using a large area blackbody integrated with a large area multi pattern target surrounded by a special shield Multi pattern target can be built using a set of different patterns However two double 4 bar patterns shown in Fig 5 8 are typically used as such multi pattern tar gets 8 The target plate is not changed during measurements and the pattern lin ear size of these two 4 bar patterns is constant However the test team can vary an gular size of the 4 bar patterns spatial frequency of the two patterns by changing distance between tested imager and test system In case of variable distance test systems it is typically acceptable to use simpli fied measurement algorithm and to carry out vertical horizontal MRTD measure ment only for positive contrast after offset correction at the beginning of the test procedure 12 Attention Stable offset must be confirmed before measurement 136 Fig 5 8 Exemplary multi pattern target black color holes white color metal plate It is possible to measure
20. 12 Kruse Paul W Uncooled Thermal Imaging Arrays Systems and Applications SPIE Vol TT51 2001 13 Manning Paul A et al Silicon foundry microbolometers the route to the mass market thermal imager Infrared Technology and Applications XXX Conference SPIE Vol 5406 2006 14 Star Safire HD QWIP thermal imaging equipment operator manual FLIR Inc P N 4100070 2006 28 3 Parameters of thermal imagers Thermal imagers generate images that can be seen by humans and it is possible to evaluate a thermal imager using human sight However it is surprisingly diffi cult even for an expert to precisely evaluate thermal imagers only by looking on im ages of typical scenery Measurement of a series of parameters is needed in order to accurately evaluate tested thermal imagers Parameters are quantitative physical measures of thermal imagers The meas urement is typically done in laboratory conditions but generally parameters of thermal imager enable an expert to predict how this imager will perform under real observation conditions Characterization of thermal imagers is relatively well standardized 1 2 15 18 19 27 There exists also a numerous literature on a subject of testing thermal imagers 3 9 12 13 14 17 21 23 25 26 31 On the basis of the mentioned above literature parameters that describe performance of thermal cameras can be in general divided into eight groups 1 Subjective image quality parameters Response parameter
21. 29 30 Saginov L Influence of spectral characteristics of IR 8 12 um photodetectors on thermovi sion system parameters Proc SPIE Vol 4340 p 305 307 16th International Conference on Photoelectronics and Night Vision Devices Anatoly M Filachev Ed 11 2000 Schwalm R Design solutions for thermal imaging devices in military vehicles Proc SPIE Vol 2736 p 259 270 Enhanced and Synthetic Vision 1996 Jacques G Verly Ed Publica tion Date 5 1996 Schwalm R Design solutions for thermal imaging devices in military vehicles Proc SPIE Vol 2736 p 259 270 Enhanced and Synthetic Vision 1996 Jacques G Verly Ed Publica tion Date 5 1996 Sousk Stephen O Shea Patrick Hodgkin Van Measurement of uncooled thermal imager noise Infrared Imaging Systems Design Analysis Modeling and Testing XVI Conference SPIE Vol 5784 p 301 308 2005 Sousk Stephen O Shea Patrick Hodgkin Van Uncertainties in the minimum resolvable tem perature difference measurement Infrared Imaging Systems Design Analysis Modeling and Testing XV Conference SPIE Vol 5407 p 1 7 2004 STANAG 4349 Measurement of minimum resolvable thermal difference MRTD of thermal cameras 1995 STANAG No 4348 Definition of nominal static range performance for image intensifier sys tems 1988 STANAG No 4351 Measurement of the minimum resolvable contrast MRC of image in tensifiers 1987 31 Thermographic Terminology Supplement 2 to Acta Th
22. 4 8 at 630 nm suggest a perfect collimator capable to project per fect images not only in infrared range but also in visible range Practically this in 84 formation is truly valid only in case of spot targets where the small spot is located exactly at the collimator focus or in case of the center point of bigger test patterns In case of low F number off axis parabolic collimators image quality of projected big targets or targets located outside the collimator focus can be much lower that the interferometric data suggest To summarize mirrors accuracy diffraction blur or collimator wavefront ac curacy should not be used as criteria of collimator quality They all can be mis leading parameters These parameters should be treated as indicators of possible collimator quality However they are not the parameters that would give warranty about quality of collimator at user hands Collimators should be characterized using a parameter called spatial resolution that depends on aberration blur and this para meter should be measured at final user facilities Such a precise condition on quality of IR collimator was proposed in Ref 5 The collimator spatial resolution V o was defined as the frequency of the smallest bar pattern projected by the collimator that the observer is able to recognize The resolution of the tested imager was defined as the Nyquist frequency vy that determines thermal imager theoretical limit It was shown in Ref 5 that to
23. Finally rules of source receiver flux calculations are presented In Chapter 2 a short review of thermal imaging technology is presented Differ ent generations of thermal imagers are discussed Some technical details of design of thermal imagers are shown Finally main applications of thermal imagers are presented Chapter 3 is devoted to definitions and measurement principles of numerous characteristics of thermal imagers At first the characteristics are divided into sever al main groups and later the definitions and each group is analyzed In Chapter 4 modern equipment for testing thermal imager is discussed At first several different types of test systems are presented Next all major mod ules of test systems for testing thermal imagers are presented and analyzed Conclu sions about requirements for these modules are shown too In Chapter 5 procedures of measurement of main characteristics of thermal im agers are discussed MRTD measurement procedure was discussed in detail due to its subjective manual characters Measurement procedures of other parameters like MTF responsivity function and noise parameters NETD FPN non uniformity 1 f 3D noise components NPSD were presented shortly due to semi automatic measurement character of their measurement and possible differences of measurement techniques using equipment from different manufacturers In Chapter 6 comments on requirements on test equipment of the Stanag 4349 stan
24. Noise generated error NGE 3 Temperature stability TS 4 Measurement spatial resolution MSR As it was discussed in Ref 4 this set of four parameters is sufficient for charac terization of measurement thermal cameras The minimum error ME is defined as a range around the output temperature Tour in which the true temperature T is located when the measurements are carried out in the conditions identical with the conditions during calibration of the thermal camera The calibration conditions exist when the tested object is a sufficiently large blackbody the distance between the tested object and the thermal camera is short in order to have negligible influence of limited transmittance of the atmo sphere environment temperature is of typical laboratory range 18 C 25 C the ob 65 ject is located in the center of the system field of view measurements are carried out for the shortest temperature span of the thermal camera and averaging the effect of a dozen or more of measurement results is used Practically the minimal error ME is an equivalent of the accuracy parameters presented in data sheets of meas urement thermal cameras 15 50 50 150 250 350 450 Tbb C Fig 3 32 Exemplary measurement error ME of several thermal cameras squares camera 1 triangles camera 2 circles camera 3 plus signs limits according to accuracy parameter The noise generated error NGE is def
25. The first case is more convenient because smaller blackbodies and smaller temper ature chambers of low power and low thermal inertia can be used In the second case bigger blackbodies and bigger temperature chambers must be used but the test conditions resemble well work conditions of real thermal imagers and bet ter effectiveness of correction of spatial noise can be achieved 110 temperature chamber black module a body WA temperature chamber thermal imager al IR FPA black body b Fig 4 24 Typical concept of correction of spatial noise From test hardware point of view two basic test modules are needed to carry out measurement of spatial noise of tested thermal imager a temperature chamber and an area blackbody The task of these two modules is to simulate real environmental conditions of work of the tested imager IR FPA module Therefore the requirements vary de pending on geographic region but let us propose requirements on these two mod ules that make them suitable for testing thermal imagers supposed to work in any climatic conditions Table 4 8 Requirements on temperature chamber Parameter Requirement Temperature range 30 C to 60 C Temperature uniformity better that 0 5 C Volume Optimized to number of imagers to be tested at the same time Temperature regulation speed Full tem
26. The top end non cooled thermal imagers offer 680x480 image resolution and are directed mostly towards more demanding military applications Non cooled thermal imagers of 320x240 image resolution are typically targeted to general sur veillance and radiometric applications security sector automotive industry non contact temperature measurements etc Imagers of 160x120 or lower resolution are targeted to mass applications in low cost intruder detection systems or as non contact imaging thermometers Technology of cooled thermal imagers is for the last decade under the pressure from non cooled technology Because of a need to use expensive cooler module the cooled technology is inherently more expensive that non cooled technology Because of this situation manufacturers of cooled thermal imagers concentrate on market of long medium range surveillance imagers or on applications that require dynamic surveillance of high speed scenarios The efforts go into four directions First reducing costs of manufacturing of II and III Gen thermal imagers of medium resolution up to 640x480 Second development of high resolution thermal im agers of image quality comparable to quality of images offered by High Definition Television minimal image resolution 1280x720 pixels is needed 9 14 Third de velopment of multi band cooled imagers capable to employ spectral phenomenon as an effective tool in both surveillance and measurement applications 4 3 Four
27. To summarize good test software should guide the user through the measure ment process and minimize possible errors This requirement can be fulfilled by semi independent software modules designed to support measurement of specific characteristic that make the user to carry out steps of the measurement algorithm and give the user precise instructions about the measurement steps Absolute mk 45 9 43 5 153 3 176 Wey swe NETD FPN nu Total NETD Total 5 66 7 SITE 30 763 dik NETO FEN NU Tead pixels coordnaies 1 x20 y 94 2175213 jel spatial frequency 1 mrad Dee E Fig 4 22 Exemplary noise window of test Fig 4 23 MTF of two different thermal software cameras 10 Typical commercial video capturing software was developed with the aim to capture and record long video sequences using as least as possible memory on hard disk Therefore such software often degrades image quality of the captured images and this degradation can influence measurement results 108 4 2 6 Optional blocks It is impossible or at least very difficult to carry out testing thermal imagers without such blocks of the test system like collimator blackbody rotary wheel set of targets PC frame grabber and test software There are also some blocks that are not strictly needed but still can be useful 4 2 6 1 Temperature chamber Thermal imagers are usually tested at l
28. Vo V3 n 5 14 where n means number of observers Fill the suitable column Correct the raw measurement MRTD using this formula 3 The latter method is recommended because in specifications manufacturers usually repeat requirements developed by military who are not interested in very short distances Therefore real minimal focus distance is usually a few times lower that value shown in specifications 138 MRTD o MRTD cor T ta 5 15 where cor 7 is the correction coefficient see Table 5 1 and ta is effective atmospheric transmittance typically equal to one Fill the suitable column When the presented above measurement procedure is carried out we get eight meas urement points and shape of MRTD curve determined accurately Four measure ment points can be still considered as satisfactory high number for typical tests If MRTD of tested imager can be roughly predicted then the number of measurement points can be reduced to three points Test results obtained using the presented measurement algorithm are shown in Table 5 11 Please note that the test results shown in Table 5 11 obtained using a variable distance test system are almost the same as test results shown in Table 5 9 Table 5 10 obtained using a variable target test system Therefore the conclusion is that if properly used then both two types of test systems generate the same test results It should be also noted that in case of measurement using the exemplary var
29. ac curately tested thermal imagers Here in this chapter we are to discuss measurement procedures of only a small but the most important group of parameters of thermal imagers MRTD MTF and noise parameters NETD FPN non uniformity 1 f noise 3D noise components NPSD The main emphasis was put on measurement of MRTD due to two basic reasons First MRTD is considered as the most important parameter of thermal imagers and it is commonly measured by both user of thermal imagers and by manufacturers of these imagers Second in spite of apparent simplicity the measurement of MRTD is difficult and it is easy to make errors in measurement procedure that can signific antly reduce accuracy of the test results It is highly probable that a test team capable to carry out accurately MRTD measurement of modern thermal imagers shall have no problems to carry out meas urement of other parameters of thermal imagers 5 1 MRTD MRTD is considered as the most important parameter of thermal imagers MRTD measurement is both simple and difficult Simple because the decisions are made by humans no tools for recording and processing electronic images are needed Difficult because many sources of possible measurement errors exist A classical subjective measurement of MRTD is a time consuming process The measurement can be shortened by using semi automatic MRTD measurement method However the latter one can be used practically only by manufacturers In ord
30. adaptation is achieved by the change of a shape of the lens ball This ability deteriorates together with aging process It is not possible to state what is exactly a spectral sensitivity range of a human eye because the spectral limits differ among the humans and the spectral sensitivity band is also on the power of incoming light However the International Lighting Commission CIE and the International Electrotechnical Commission CIE 15 de termined a relative spectral sensitivity function of a human eye in the spectral range from 0 38 um to 0 78 um and these values are often considered as limits of the vis ible range If we compare a spectral sensitivity curve of a human eye with a characteristic of spectral emission of light emitted by the Sun or reflected by the Moon then we will notice that a human eye is very well adapted for illumination conditions on the Earth where the light emitted by the Sun is the dominant source of light Fig 1 4 0 8 0 4 0 2 0 0 A ym Fig 1 4 Relative spectral sensitivity of human eye Horizontal field of view of a human eye is about 40 and vertical field of view is about 30 Due to its ability to rotate a human eye can effectively see detection of movement almost in full hemisphere However good sharpness of an image is achieved only in the central 10 field of view the best sharpness in 2 field of view Due to aberrations of the lens of the human eye and diffraction phenomen
31. and the results are more stable Therefore it is highly recommended to carry out MRTD measurement only by test teams employing people who are so called qualified MRTD observers The latter ones are the people having no medical problems with their eyes and who passed special training where they became familiar with MRTD measurement They made exemplary MRTD tests for at least dozen or more hours and then repeatability of their test results and their ability to detect and recognize low contrast targets in noisy environment was confirmed The training of qualified MRTD observers should be done using a series of different real market thermal imagers Such training can be also supported using computer simulators During real work the observation time is quite often limited The observer must hurry up to detect and recognize the target he is looking for in the scenery However it is commonly accepted that during MRTD tests the observer is allowed unlimited in reasonable range viewing time It improves little test results in comparison to the limited time test conditions 5 1 2 Detection criterion It is generally accepted in literature that the observer can consider that he sees the 4 bar pattern when he is able to count four separate bars The bars do not to be seen all the time They can be covered by temporal and spatial noise but still the ob server must be sure that he can recognize all four bars The bars can be distorted One or t
32. be measured horizontal if vertical MTF is to be measured Some manufacturers use also tilted MTF measurement method when the edge target is not horizontal or vertical but tilted by 45 This method can increase measurement accuracy in case when blurring in the analyzed image is very small smaller than about 3 5 pixels Setting proper settings of thermal imager typically Gain to minimal Level to average Regulation temperature difference differential mode using control soft ware until a clear image of the edge target can be seen Typically differen tial temperature in the range from about 2 K to 8 K is used The contrast of the target should be moderate It should be avoided situation when bright semi moon target is very white and background is very dark The situation when the contrast of the target and background is low should be also avoided Generally the aim is to get situation when both the target temper ature and the background temperature are in the linear range of SiTF func tion of the tested imager Some manufacturers offer software tools that en able easy checking of brightness of the target and the background to speed up a process of finding proper contrast of the image of the targets Situation when we have brightness of the target in the range from about 120 to 150 grey levels brightness of the background in the range from about 60 to 90 can be treated as a safe solution Focusing of tested thermal imager to get ma
33. case of variable distance test systems the target tem perature can be measured directly or at least the temperature sensor has a very good thermal contact with the target plate Therefore the offset of variable distance test systems is much smaller than in case of variable target test systems Next the offset is more stable if measurements are done at relatively stable ambient temperature the system is not exposed to sun wind ambient temperature changes are slow 78 There are also several serious disadvantages of variable distance test systems First large blackbodies and large targets and needed This makes speed of regula tion of blackbody temperature significantly lower than in case of smaller blackbod ies used by variable target test systems There also technical problems to assure re quired high thermal uniformity and temporal stability of bigger blackbodies Second large rooms or corridors are needed if the tests are to be carried out in in door conditions It is not so easy to find a 50 m long corridor that is sometimes needed to test a surveillance thermal imager Third the test system requires some kind of protection against environmental condition wind rain sun if they are to be used at outdoor conditions Fourth expanded tests measurement of a long series of parameters of thermal imagers can be done using the variable distance test systems but the test procedures are not as user friendly as in case of the variable target t
34. cea tat osed cut enteeeenee ts decease 116 SL MART D D AE ootedssessncsecsccsvaestdcasesdessesscevapdavtessacdscacensdavectectescontestestansussaseass 116 Dell Test tedni arse a a Seheeese a Sie tees te a ede sdeth ssh a ae Teach 117 5 1 2 Det ction THET ON ipanen e aaah ea teeta nara eae 117 5 13 Test Cnvironinient lt 2 ss cts seeks Airave die Hive a e e a ed Mes 118 SCART APOGIS 055 55 dsectbatedvascduntad Zee de nnsva deh cayue E Saad ecoeg E E E A E 120 Deh 5 Onset ia ne ieee ae ae eee epee esd 124 RD imager Setting face ses c 5s coe esec sees head cae cdesees E E T 125 Sul 7 Static dynamic image S oerien aeia eects oles setae neces ceded RAN kk 126 5 1 8 Differential temperature Steps cceceeccecceseeseceeceseceeceseeseeeseceeeeseecsaeeeeseeeee 126 5 1 9 Measurement methods osiinsa ia i a 127 5 1 10 SoftWare SUppotis nenene a a E E a a en 140 5 1 11 Interpretation of MRTD measurement results ccecceeseesseeteeseeeeeeeeeeeee 141 52 MAL Bincsccectechsaceredecsectuessdecesesiasocedvcossansusdacestescdessessectusasdbeedcaceuasesteatececacsedz soeesoadsene 143 De2e ly MGasurement CONn Cpt ireccrze sss s0rs cusicevestes iesesevovrestacseantvaesdavehgend s EEE 143 5 2 2 Measurement procedure cccccccescesceeseesceessesceeseceseeseceeeeseeeseceeeeseeeaeeeseneeeee 145 5 2 3 Interpretation of measurement results cccesceeceeseeeceeteeeeeeteeeeeeneeeeeeeeees 146 5 3 Responsivity function SiTF
35. color of the incoming light A human eye is characterized by a certain temporal inertia of about 0 1 0 2 s It means that if a sequence of still images is shown in rapid succession the brain will reassemble the still images into a single moving scene Therefore this phe nomenon enabled development of television and computer animation but at the same time temporal inertia of a human eye is a significant handicap of human sight when fast phenomena are to be observed The eye can perceive images of maximal contrast not higher than about 100 1 However the eye can adapt to different illumination conditions by readjusting its exposure both chemically and by adjusting the iris Several seconds are needed for initial dark adaptation Full adaptation through adjustments in retinal chemistry can take as much as about thirty minutes However if we consider the possibility of a human eye to perceive images at both very bright day and dark night then we can say that after full adaptation a human eye can achieve dynamic contrast ratio of about 1000000 1 The process is nonlinear and multifaceted An interruption by light nearly starts the adaptation process over again To summarize a human imaging system can be considered as a small size uni versal electro optical imaging device of very high capabilities that enables human perceiving high quality images of the neighbor scenery at a variable level of illu mination The human eye has much better dynamic r
36. contact between the targets and the rotary wheel then difference between target temperature and wheel tem perature is negligible and target temperature can be indirectly measured 100 Both the target plates and the rotary wheel should be put inside an enclosure that would reduce influence of conditions in a test room on temperature of the wheel with the targets We must keep in mind that variations of ambient tem perature in test rooms are often at level of several Kelvins in a period of a dozen minutes If similar variations of temperature of the target wheel occur then accurate measurement of parameters of thermal imagers is not possible To summarize we can present the following requirements on the rotary wheels 1 Manufactured from high thermal conductivity material 2 Good thermal contact between targets and the wheel where the targets are fixed 3 Preferably motorized type 4 Easy and speedy target exchange 5 The targets should be protected against influence of external conditions on their temperature 6 Positioning repeatability at level about 1 mm typical tests or about 0 1 mm automatic targets recognition tests 4 2 4 Targets Targets for testing classical visible near infrared imaging systems are manufac tured by creating opaque or semi transparent coatings on transparent glass substrate When a diffuse light source is put behind the target then the tested visible cam era sees a target formed by the coating o
37. effectively testing of modern thermal imagers Second some important questions about test methods and precise requirements on test equipment are not answered or it is difficult to find a proper answer in numerous literature on subject of testing thermal imagers Third freely detailed technical literature available from manufacturers of equipment for testing thermal imagers is rare and sometimes compiled in a way to show superiority of their own equipment over the equipment offered by competitors 14 This book consists of seven chapters Chapter 1 is already read introduction of this book where the concept of electro optical imaging systems human sign properties division of electro optical imaging systems terminology of E O sys tems basic metrological terminology were introduced Thermal imagers employ the phenomenon of thermal radiation create thermal image of the scenery being observed Therefore the whole Chapter 2 Thermal ra diation is devoted to the discussion about properties of thermal radiation Firstly quantities and units of this kind of radiation are presented Next basic laws are dis cussed The laws describe only a phenomenon of thermal radiation emitted by an ideal type of objects i e blackbodies Therefore radiant properties of real mater ials are also discussed to enable us the analysis of radiation emitted by real materi als Further on the influence of the atmosphere on propagating radiation is dis cussed
38. human diversity only approximate limits were given Further on what is the most important these recommendations are not accepted in most communities working in the field of optical radiation due to many mostly historical reasons Table 1 2 Division of optical radiation recommended by the CIE Confusion in area of limits and further division of sub ranges of optical radi ation is particularly clear in case of infrared radiation range A dozen or more pro posals on division of infrared range have been published in the literature Precise di vision of the infrared radiation is particularly important for any book on thermal im agers Therefore for the purpose of this book a precise division of infrared radiation shown in Table 1 3 will be used 10 The division shown in Table 1 3 is based on the limits of spectral bands of commonly used infrared detectors Wavelength of 1 um is a sensitivity limit of popular Si detectors used in TV imagers Similarly 3 um wavelength is a long wave sensitivity limit of PbS and InGaAs detectors used in near SW IR imagers Wavelength of 6 um is a sensitivity limit of InSb PbSe PtSi detectors and HgCdTe detectors optimized for 3 5 um atmospheric window used in medium wave thermal imagers Finally 15 um wavelength is a long wave sensitivity limit of HgCdTe de tectors optimized for 8 12 um atmospheric window or non cooled detectors with long wave filter used in long wave thermal imagers Table 1 3 Divisi
39. many years The US common module HgCdTe arrays that employ 60 120 or 180 photoconductive elements are the prime example of Gen 1 thermal cameras Fig 2 2 Exemplary Gen 1 thermal camera LORIS courtesy of FLIR Inc Scanning cameras built using linear or 2D focal plane arrays FPA of elements number higher than about 100 but lower than about 10000 are the Gen 2 thermal cameras Temperature resolution NETD of these cameras is improved up to the level of about 0 1 K They are also characterized by smaller weight and size and improved reliability The 1980s is a period when most modern army forces started to use the second generation thermal cameras The cameras of this generation are presently majority of all military thermal cameras New version of these FPAs offered in a form of a single chip fully integrated with readout electronic are even now an attractive solution for many observation applications Thermal cameras built using these improved linear FPAs are often termed Gen 2 Temperature resolution NETD of Gen 2 can be improved up to the level of about 0 05 K Typical ex amples of these systems are HgCdTe multilinear 288x4 arrays fabricated by So fradir both for 3 5 um and 8 10 5 um bands with signal processing in the focal plane photocurrent integration skimming partitioning TDI function output preamplification and some others Third generation cameras are non scanning thermal
40. of 10 C The accuracy must be 0 5 generally and 0 01 C for ATs between 0 C and 2 C 5 2 4 Any variation in temperature across the useful area of the test pattern shall be so small as not to be detectable on the im age by the observer 5 2 5 The measurement temperature is that of the test pattern which should be 20 2 C unless otherwise specified 5 3 The transmission losses between the test pattern and the imager and the emissivities of the test pattern and the black body should be known and taken into account when results are calculated 158 collimator Relative uncertainty of spatial frequency of 4 bar targets manufactured by Inframet is 4 Uncertainty of determination of focal length of IR collimators offered by most manufacturers of test equipment is below 1 Emissivity of TCB series black bodies is 0 97 0 01 Emissivity of typical IR targets offered on the market is 0 97 0 01 sometimes gt 0 98 Typical differential temperature range for blackbodies used in test systems is 25 C and 75 C much wider than required Dif ferential accuracy of blackbodies is not worse than 0 01 C at entire temperature range 10 C Non uniformity of blackbodies for temperature differences below 5 C is typically below 10 mK and is non detectable for thermal im agers This is an requirement to be ful filled by users of test equipment not requirement on the test equip ment However this requirement can be e
41. of 3D array Nryy Fig 3 17 The 7 dimension represents time or the framing sequence The H dimension and V dimension give spatial information In case of staring systems m and n in dicate the pixel location in case of scanning systems m refers to a pixel location but refer to time or sample number in a digitized analog signal Names of each components and information provided by each component are presented in Table 3 5 The noise components are calculated by converting the 3D array into a series of 2D or 1D arrays The conversion formulas are presented in Table 3 6 Fig 3 17 Three dimensional noise model coordinate system m row number n column number N frame number It is possible to get detail information about the nature of a noise phenomenon by the analysis of a sequence of the images generated by the tested thermal camera using the 3D noise model and this model is often used by manufactures of thermal cameras On the other hand the 3D noise model due to long series of parameters is complicated and this model is very rarely used by users of thermal cameras who prefer simple solutions Because of these customer demands for simplicity even manufacturers rarely publish data using the 3D noise model 45 Table 3 5 Noise components of the 3 D noise model 3D com Number Comments Information ponent of ele ments 1 Svar mxnxN Random 3 D noise 2
42. of A 8 can be considered as ac ceptable Therefore we can conclude that if the collimator is to be used at highly variable ambient temperature conditions field conditions then the collimators mir ror should be made from Zerodur or at least from Pyrex In case of laboratory con ditions when the ambient temperature is almost stable then all optical materials mentioned earlier are acceptable but due to lower cost mirrors made from Pyrex or from optical crown glass should be preferred 91 Table 4 5 Comparison of four materials used for mirror substrates Material Coefficient of Typical Typical surface Cost thermal cosmetic accuracy expansion C surface quality Optical crown 10 80 50 N4 A 12 low glass commonly BK7 type Pyrex 5 10 60 40 A 4 A 10 moderate Synthetic fused 8 107 60 40 A 10 A 20 high modera silica te Zerodur 4107 60 40 up to A 20 high 4 2 2 Blackbody Blackbody is an ideal body that completely absorbs whole radiant energy strik ing it and therefore appears perfectly black at all wavelengths The radiation emit ted by such a body when heated is referred to as blackbody radiation A perfect blackbody has an emissivity equal to unity Emissivity of real technical blackbodies is close to unity There are generally two methods to design technical blackbodies of emissivity nearly equal to unity First method is to manufacture a cavity in a block of material of high therm
43. of brightness of pixels of images generated by thermal cameras In other words the 1 f noise creates slow temporal changes of spatial noises FPN and non uniform ity 1 f noise has directly no influence on image quality perceived by a human ob server Therefore 1 f noise is typically omitted in noise analysis of thermal cameras and this parameter is rarely measured NETD FPN and non uniformity are con sidered as a basic trio noise parameters and are commonly measured NETD 1 Capturing a short video sequence of thermal images generated by the tested thermal imager if the number of captured video frames is no 1 f noise C 3 8 52 FPN U more than about one hundred we can assume that 1 f effect is negligible 3D noise array is created NETD ofa single pixel is calculated as a standard deviation of temporal variation of intensity of this pixel with time NETD of an analyzed group of pixels or the whole thermal image is cal culated as an average NETD ofall pixels included in the analyzed group Calculated NETD in digital level units is converted to the NETD in tem perature units _ NETD digital levels NETD C a 3 9 Capturing a short video sequence of thermal images generated by the tested thermal imager Averaging operation of the captured frames A group of frames is replaced by a single frame temporal noise component is eliminated or at least re duced High pass frequency filtering operation o
44. of modern undersampled thermal imagers 29 MTDP is a temperature difference at which four three or two bars can be resolved by an observer with the test pattern at the optimum phase position Using MTDP concept imager performance over Nyquist frequency can be still analyzed and a significant drawback of classical MRTD is eliminated At the same time MTDP uses much of MRTD concept the same 4 bar target the same test equip ment and it is easy for people familiar with MRTD concept to understand and ac cept MTDP concept Next MTDP concept was implemented in TRM3 a well know model for performance evaluation of thermal imagers 30 Therefore MTDP has a real chance to become in future the main figure of merit of thermal imagers if gets support from any international standard Fig 3 6 Image of afourbar target during MTDP measurement 3 1 2 MDTD The MDTD is a subjective parameter that describes ability of the imager human system for detection of small size targets It is a function of a minimum temperature difference between the circular target and the background required to detect the tar get by an observer versus inverse spatial size of the target The MDTD is measured by determining the minimum temperature difference between the target and the background required to detect the thermal image of a tar get for targets of different spatial dimensions Fig 3 7 Images of two circular targets of different angular sizes The measure
45. parameters response parameters noise parameters image resolution parameters and geometric parameters give us general information about performance of thermal cameras in surveillance applications Accuracy para meters are vital to evaluate thermal cameras for measurement applications Opera tional parameters provide information about some practical aspects of work with both types of thermal cameras Table 3 1 Performance parameters of thermal cameras No Subjective image quality paramet ers Response para meters Noise parameters Image resolution para meters 1 MRTD minimum resolvable temper ature difference Responsivity func tion NETD noise equivalent temper ature difference Number of pixels lines IFOV instantaneous field of view DAS detector angular substance 2 MDTD minimum detectable temper ature difference SiTF signal trans fer function FPN fixed pattern noise MTF modulation trans fer function 3 Auto MRTD Dynamic range Saturation level Non uniformity CTF contrast transfer function EIFOV effective in stantaneous field of view limiting resolution ence perceived 4 TOD triangle ori SRF slit response 1 f noise PVF point visibility entation discrimin function factor ation 5 MTDP minimum ATF aperiodic 3D noise model measurement spatial temperature differ transfer function nine
46. project images of targets fixed to a rotary wheel using an reflective collimator as an image projector The tested thermal imager is located at the output of IR collimator and the target is loc ated at the collimator input the focal plane The distance between the target and the tested imagers is very short The distance is typically no more than about 4 meters if we analize the optical ray way and the distance is usually not within the focusing range of typical surveillance thermal imagers However due to use of the collimator as an image projector the imager sees the target as a very long distance object that is within imager focusing range Next a series of targets is fixed to rotary wheel By rotating the wheel it is possible to exchange quickly tar gets By changing target dimensions the changes of distance are simulated IR collimator Optical table Tested E imager j i frame i i n grabber a T Fig 4 2 Diagram of the variable target test system image projector 74 CTCB Controller Fig 4 3 Photo of the DT 1500 variable target test system courtesy of Inframet www in framet com The variable distance test systems Fig 4 4 Fig 4 5 generate a thermal image at the plane of the target fixed to a large area blackbody The tested thermal imager sees this image directly and generates a distorted copy of the image generated by the test system Because the variable distance systems do n
47. silicon CCD CMOS technology At the same time SWIR imagers have proved to be useful in telecommunication sector en abling accurate coupling of optical fibers working at 1 53 um and in museums for painting reflectography 26 2 5 Technology trends Thermal imaging is one of the technologies of paramount importance for milit ary amp security sector Thermal imaging has found also numerous applications in a ci vilian sector Therefore it is not strange that there is a lot of efforts to improve exist ing technologies of manufacturing thermal imagers and to develop new technolo gies We can distinguish several trends in thermal imaging technology 1 Low cost low medium resolution non cooled thermal imagers 2 High resolution cooled thermal imagers of improved surveillance capabilit ies 3 Dual band thermal imagers 4 Multi sensor systems Technology of non cooled thermal imaging experienced very rapid growth dur ing the last decade 13 Parameters of non cooled imagers improved so much that nowadays non cooled imagers dominate on the market of short range surveillance thermal imagers in both military and civilian applications The critical factor on this market is a price Therefore now the technology efforts concentrate on decrease in manufacturing costs but still keeping or even improving the image quality and reli ability Critical areas are two modules of non cooled thermal imagers infrared fo cal plane area and infrared optics
48. temperature difference AZ when the observer can recognize target Calculate the offset as offset AT AT_ 2 5 10 Attention Both AT or AT can be positive or negative numbers Insert the offset value to blackbody software memory please refer to blackbody operational manual Blackbody indications are corrected and offset influence is eliminated 3 Increase slowly see recommended steps of differential temperature showed in Table 5 5 positive temperature difference until observer I starts to recognize the vertical 4 bar pattern Write the current temperature difference value as horizontal AT to test results table 4 Decrease slightly differential temperature until both patterns disappear and then increase it slowly until observer I starts to recognize horizontal 4 bar pattern Write the current temperature difference value as vertical AT to test results table Exchange observers 6 Repeat the previous two points until all observers carried out MRTD meas urements for the current target spatial frequency point 7 Exchange targets using the rotary wheel for a new target of higher spatial frequency Nn 134 Repeat points 3 6 until MRTD measurement was done for all targets Calculate MRTD values for each observer at one spatial frequency as the recorded positive temperature differences AT MRTID AT Fill the columns of the test results table Calculate raw measurement MRTD as an average MRTD from each ob
49. that show bigger blurring for such temperature difference due to crosstalk effect than for lower temperature difference below 2 K For such imagers it is necessary to use medium Gain level Original MTF WTF 3 8 8 8 8 8 8 P a inai k e O ee oe ee ee E Sostis feauency 1 mrad Is Fig 5 13 MTF of the same thermal imager a Edge enhancement turn off b Edge enhancement turn on 5 3 Responsivity function SiTF Responsivity function is a function of an output signal screen luminance or electrical signal versus target temperature absolute or relative in case of large constant size target The responsivity function is usually S shaped The responsivity function depends strongly on imager settings like level or gain There is no a single responsivity function that could characterize fully even simple thermal im agers Measurement of at least several functions is need to get precise information about imager performance at different combinations of gain and level settings The responsivity function can be also presented in simpler form of three digit al parameters SiTF saturation level and dynamic range that are determined on the basis of measurement results of the responsivity function The signal transfer function SiTF or the responsivity is the most important parameter from the set of parameters mentioned earlier 147 The signal transfer function SiTF is determined on the basis of the linear part o
50. that require only relative temperature measurement and the applications that require absolute temperature measurement Although the same cameras can be typically used in both applications there are different cri teria of assessment of camera suitability for these two groups of applications If the camera is used in applications when only relative temperature measurement is needed like in non destructive thermal testing NDTT then the quality of the thermal image of the tested object is usually the most important criterion like in the case of the surveillance thermal cameras If the measurement thermal camera is used in applications when an absolute temperature measurement is needed then accuracy of temperature measurement results is the most important criterion 19 2 3 Spectral band Objects of typical earth temperatures emit radiation mostly in the spectral region from about 3 um to about 15 um Thermal radiation emitted by these objects dom inate over the radiation reflected by them at this spectral range because the radiation emitted by sun moon stars and typical artificial sources is weak for wavelengths over 3 um There are two atmospheric windows in the above mentioned range the 3 5 um window and the 8 12 um window Therefore there are two main types of thermal imaging systems the middle wave MW systems using the 3 5 um win dow and the long wave LW systems using the 8 12 um window and rarely avail able commercially SW systems
51. the blackbody controller to 2 C 129 11 12 14 15 20 21 22 23 24 25 Put target of the lowest spatial frequency in the active position the position when target is located at collimator focal plane and can be seen by the tested thermal imager Focus the imager until you get a sharp image of the 4 bar pattern Rotate the imager until the image of the 4 bar pattern is located in the center of imager field of view Set using the blackbody controller temperature difference equal to zero Increase slowly see recommended steps of differential temperature showed in Table 5 5 positive temperature difference until observer I starts to recognize 4 bar patter Write the current temperature difference value as AT to Table 5 7 Decrease slightly the positive temperature difference until the 4 bar pattern disappears Exchange observers Increase slowly positive temperature difference until observer II starts to recognize 4 bar pattern Write the current temperature difference value as AT to column 7 of Table 5 7 Decrease slightly the positive temperature difference until the 4 bar pattern disappears Exchange observers Increase slowly positive temperature difference until observer III starts to recognize 4 bar pattern Write the current temperature difference value as AT the Table 5 7 Decrease temperature until the observer I starts seeing negative 4 bar target Typically it is needed
52. the possibility of estimation of performance deterioration of tested thermal imagers in order to predict their life time and to create optimal plan of repairing these imagers The typical test level is also recommended for manufacturers of thermal imagers who are looking for a test system needed both to verify final quality of manufactured imagers and to get information needed to optimize manufacturing of thermal imagers at production line The expanded test level is recommended generally for advanced manufacturers or R amp D teams who want to get very detail information about tested thermal imagers in order to make design improvements This test level is particularly recommended if the tested imager is to be used not only for classical surveillance task surveillance using human observers but also as a module of automatic target recognition system 3 7 References 1 ASTM standard E 1213 2002 Standard Test Method for Minimum Resolvable Temperature Difference for Thermal Imaging Systems 2 ASTM standard E 1311 99 Standard Test Method for Minimum Detectable Temperature Dif ference for Thermal Imaging Systems 3 Bijl Piet Valeton J Guidelines for accurate TOD measurement Infrared Imaging Systems Design Analysis Modeling and Testing X SPIE Vol 3701 14 25 1999 4 Chrzanowski K Evaluation of commercial thermal cameras in quality systems Optical En gineering Vol 41 No 10 2002 70 5 Chrzanowski K Park S N E
53. the tested thermal camera It is recommended that the collimator aperture should be at least 10 bigger than the diameter of the optics of the tested thermal camera If this condition is not fulfilled then additional errors of measurement of parameters of thermal cameras are generated Diameters of optics used in modern thermal cameras varies greatly from 10 30 mm diameter objectives typically used in commercial thermal cameras up to 200 250 mm diameter objectives used in some ultra long range surveillance thermal cameras Therefore only big collimators of apertures of about 300 mm can be used for testing all thermal cameras available on the market However bigger collimator means also more expensive more bulky instrument At the same time we should remember that thermal cameras of optics diameter of over 130 mm are very rarely met no more than about 1 of the market probably much less We can expect also some problems when testing thermal cameras of wide field of view with small optics like commercial thermal cameras using big collimator of long focal length Even the biggest targets can be small for the tested imager and measurement of MRTD at low spatial frequency cannot be carried out Therefore it seems that collimators of aperture of about 150 200 mm represent an optimal choice Such collimators enable testing almost all thermal cameras available on the market In case when test area is limited to cameras of small aperture then smaller ap
54. the tests of thermal imagers are done by analysis of analog video signal using oscilloscopes However this method of measurement of noise parameters is rarely used and shall not be discussed later 5 4 2 Measurement procedure Measurement of noise parameters of thermal imagers is typically carried out ac cording to the algorithm presented below 1 Regulation of settings of the tested imager If thermal imagers were truly linear imaging systems then noise parameters would not depend on imager settings like gain or brightness level However thermal imagers are only quasi linear systems and measured noise parameters depend on setting of the tested imager The tests are typically done for the fol lowing settings Gain maximal Brightness level medium However it is not arule In any case the imagers settings Gain Level Image Enhancement should be written down for future analysis Next it is critical for successful measurement of noise parameters that the im ager setting like Gain Level should be the same as settings used during measure ment of SiTF Further on it is strictly forbidden to use Auto Gain mode during noise measure ments 152 2 Regulation of angular position of the tested imager relative to the uniform target or inverse As it was discussed in previous section different methods to generate uniform image are used Therefore position of the tested imager relative to the uniform target blackbody can be different Versi
55. to measure MRTD proportions cut in characteristic metal sheet it is necessary to rotate targets to measure both vertical and horizontal MRTD 2 pinhole circular pattern cut MDTD a set of targets with various size is target in metal sheet needed to measure MDTD characteristic small pinhole target is needed during PVF measurement 3 square single square ATF or a set of square targets for ATF target pattern cut in metal SiTF measurement sheet NETD single square target for SiTF NETD FPN FPN measurement old technique option 4 slit target single long slit cut SRF MTF aset of slit targets of different width is in metal sheet needed to measure SRF single very narrow slit target for MTF measurement 5 interlace single long narrow scanning needed to check interlace scanning target slit pattern cut in adjustment adjustment as well as to identify strapped metal sheet skewed dead or dead channels in scanning imagers by 45 channels These effects appear in a form of deviations from the ideal smooth diagonal line 6 edge target a half moon of ESF MTF ESF edge spread function is measured sharp smooth edge directly MTF is calculated on the basis of pattern cut in metal measured ESF sheet 7 alignment patterns of pinhole boresightin different sizes are needed depending on target and cross g focusing imager field of view combinations and alignment 8 double 4 double 4 bar MRTD it is p
56. until the gas is liquefied as it leaves 25 the throttle valve Because Joule Thomson coolers require the supply of pressurized gas they are rarely used in thermal cameras but they are typically used in IR guided seekers where the required working time is relatively short Both Stirling coolers and Joule Thomson coolers are relatively expensive com ponents that represent a significant portion of cost of a whole thermal camera Therefore it was highly desirable to eliminate these components as it has been done recently by introduction of non cooled FPA based on microbolometer and pyroelec tric ferroelectric technologies However please note that so called non cooled FPAs usually require temperature stabilization and thermoelectric coolers are usually used in the non cooled thermal cameras The thermoelectric coolers employ the effect of Peltier that makes possible to generate the temperature changes using current flows in a circuit consisting of two dissimilar conductors A big disadvantage of the thermoelectric coolers is their non ability to cool detectors down to very low temperatures temperature difference of not more than about minus 50 70 C relative to ambient temperature can be achieved However low cost of these coolers is their big advantage in sharp con trast to expensive Stirling coolers and Joule Thomson coolers Apart from the MWIR thermal cameras and the LWIR thermal cameras there are also SWIR cameras of a spectral band locat
57. using two presented earlier formulas are usually quite small below 2 Now we will present the methods that can be used to measure all four noise components 1 f noise 1 Capturing several tens of group of frames separated by a long time inter val Each group is treated as a 3D noise array There are n such 3D noise arrays where n is the number of captured groups of frames 2 Averaging operation of frames within the each group A group of frames 3D noise array is replaced by a single averaged frame High frequency temporal noise component is eliminated or at least reduced New 3D array is created that carried information only about low frequency temporal noise 3 Calculations of 1 fnoise of a single pixel as a standard deviation of tempor al variation of intensity of this pixel 4 1 fnoise of an analyzed group of pixels or the whole thermal image is calculated as an average of 1 fnoise of all pixels included into the analyzed group 5 Calculated 1 f noise in digital level units is converted into the 1 f noise in temperature units _ 1 f noise digital levels SiTF 6 I f noise can be also expressed as a percentage of average signal of the analyzed area or as a percentage of NETD 1 f noise component is noticeable only if we analyze temporal trends in frames of a long video sequence say at least a few minutes The 1 f effect is not noticeable in short video sequences Generally 1 f noise creates an effect of slow variations
58. 0 31 32 33 34 35 Holst G C The Infrared amp Electro Optical Systems Handbook Vol 3 Electro Optical System Design Analysis and Testing Chapt 4 pp 206 207 SPIE 1993 Holst G C Infrared Imaging System Testing Vol 4 Chapt 4 in The Infrared amp Electro Optical Systems Handbook Michael C Dudzik ed SPIE 1993 International Lighting Vocabulary CIE Publ No 1 7 4 IEC Publ No 50 845 1987 ISO 15529 Principles of measurement of modulation transfer function MTF of sampled imaging systems 1999 Li X An investigation into the stability of industrial platinum resistance thermometer Hart Scientific Lloyd J M The Infrared amp Electro Optical Systems Handbook Vol 3 Electro Optical System Design Analysis and Testing Chapt 1 SPIE 1993 MIL STD 1859 Thermal Imaging Devices Performance Parameters Of 1983 Driggers R Van Hodgkin A Vollmerhausen R O Shea P Minimum resolvable temperature difference measurements on undersampled imagers Proc SPIE Vol 5076 Infrared Imaging Systems Design Analysis Modeling and Testing XTV 2003 STANAG 4349 Measurement of minimum resolvable thermal difference MRTD of thermal cameras 1995 Tavener J P Common errors in industrial temperature measurement Isotech Journal of Thermometry Vol 6 No 2 1995 20 27 The Photonics Handbook Book 3 p H398 H404 Laurin Publishing Co 1993 Walker R Achieving 0 25mK uncertainty with an integrated circuit resi
59. 0 09 0 15 0 12 0 1 0 09 3 0 67 1 50 0 15 0 21 0 18 0 19 0 25 0 22 0 27 0 33 0 3 0 23 0 21 4 0 50 2 00 0 32 0 38 0 35 0 37 0 43 0 4 0 52 0 43 0 47 0 43 0 4 510 40 2 50 0 62 0 68 0 65 0 67 0 62 0 64 0 77 0 73 0 75 0 72 0 66 610 33 3 00 1 07 1 13 1 1 1 37 1 43 1 4 1 37 1 43 1 4 1 3 1 2 710 29 3 51 3 47 3 53 3 5 3 17 3 23 3 2 3 37 3 43 3 4 3 37 3 1 1025 4 00l J o m e ile fle a bar width in mm v MRTD in C units spatial frequency in lp mrad AT temperature difference in C units The measurement algorithm presented earlier generated two significantly differ ent MRTD curves horizontal MRTD and vertical MRTD This is a typical situation for scanning thermal imagers horizontal MRTD significantly better than vertical MRTD particularly in high frequency range the differences between horizontal MRTD and vertical MRTD are typically small sometimes negligible There is also no rule which MRTD is better for staring im agers 132 In case of staring thermal imagers 2 D w MRTD deg C o 0 02 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 frequency 1 mrad Fig 5 7 Exemplary horizontal MRTD and vertical MRTD measurement results squares hori zontal MRTD triangles vertical MRTD no symbols average MRTD There is a question of big importance what to do when there is a difference between ho
60. 2 2 5 5 74 13 7 9 25 9 10 45 1 125 27 1 35 3 375 45 2 25 5 625 76 13 8 9 5 10 10 5 1 25 28 1 4 13 5 46 23 5 75 78 13 9 9 75 11 10 55 1 375 29 1 45 13 625 47 12 35 15 875 80 4 10 12 10 6 1 5 30 1 5 3 75 48 12 4 6 82 14 1 10 25 13 10 65 1 625 31 1 55 3 875 49 2 45 6 125 84 14 2 10 5 14 0 7 1 75 32 1 6 14 50 12 5 6 25 86 4 3 10 75 15 10 75 1 875 33 1 65 4 125 52 2 6 6 5 88 4 4 11 16 10 8 32 34 1 7 4 25 54 2 7 6 75 90 14 5 11 25 17 10 85 2 125 35 1 75 4 375 56 2 8 7 95 14 75 11 875 18 10 9 12 25 36 1 8 14 5 58 12 9 7 25 100 5 12 5 R distance a 10 mm width of a single bar of the bigger pattern a 4 mm width ofa single bar of the smaller pattern Multiple 4 bar targets can speed up measurement process even more However the use of these targets is more risky due to not verified influence on measurement accuracy in situation when the observer sees several patterns at the same time Next parabolic off axis collimator are theoretically aberration free but only for on axis point Images of 4 bar patters located outside the target center can be distorted noticeably in case of collimator of low F number Therefore the multiple 4 bar tar gets should be used only when short measurement time is critical Fig 5 3 Multiple 4 bar target 123 Rotation of the pattern bars always influence test results The worst results we can get when the rotation angle is about 45 However
61. 3 1 The list of parameters commonly measured during tests of thermal imagers is much shorter than list of parameters discussed in this chapter due to several reasons First MRTD gives information about both temperature sensitivity of tested thermal imagers and about its spatial resolution At the same time MRTD is the only parameter of thermal imagers presented in several internationally recognized test standards Therefore there are some test teams that limits tests of surveillance thermal imagers to measurement of only this parameter Other subjective image quality parameters like MDTD TOD MDTP are measured rarely 68 Second measurements of response parameters are typically limited to measurement of SiTF parameter because it is the only response parameter that is needed to carry out measurement of important noise parameters Third nine components of 3D noise model give the most detail information about noise present in images generated by tested thermal imagers However the concept of 3D noise model is not simple and the information provided by this parameter is not easy for interpretation Therefore almost all users of thermal imagers and majority of manufacturer test teams prefer measurement of more simple parameters like NETD FPN non uniformity Fourth a long series of image resolution parameters are MTF related parameters If MTF function is measured we can determine almost all other image resolution parameters Therefore it is log
62. 5 is the simplest and the cheapest However temperature distri bution on surface of a such a ad hoc blackbody is not uniform At the same time the temperature distribution is not temporarily stable Therefore this method gener ates often non stable test results particularly spatial noise and should be avoided One of earlier discussed methods to generate uniform image should be used All thermal imagers of the past and majority of thermal imagers present on the market generate images in form of an analog video signal PAL NTSC standard This means that practically the output image is generated as analog electrical signal Therefore in the past the noise of thermal imagers was almost exclusively meas ured using typical measuring tools for analog electrical signals the oscilloscopes The measurements results were obtained in Volts units and later converted into temperature units Frame grabber cards for modern PC units give easy possibility to convert output analog signal from tested thermal imagers into a digital image Analysis of digital images using modern computer technology is much easier than analysis of analog electrical signals using oscilloscopes although there are also big improvements in digital oscilloscopes Therefore nowadays noise parameters are usually determined by analysis of sequence of digital images generated by frame grabbers However we should remember that there are still laboratories mostly manufacturers labs where
63. Analysis of only a sequence of images of uniform target can give information about imager noise in gray level units or in volts in analog electrical signal is ana lysed not in temperature units as required The conversion of noise in gray level units to temperature units can be done only when SiTF is known SiTF gives in formation about relationship between input temperature of the target and output brightness of image of this target Therefore in order to determine noise paramet ers we must first measure responsivity function or SiTF as described in the previ ous section Sometimes measurement of SiTF is treated as part of noise measure ment procedure and SiTF is presented together with noise parameters Manufacturers of equipment for testing thermal imagers use different measure ment procedures to measure noise parameters These differences can generate some differences in test results The differences can be divided into two groups A Method of generation of uniform image B Type of output signal generated by the tested imager to be analysed analog video signal or digital signal Uniform image needed for noise measurement can be generated using several methods 1 Thermal imager is looking into a small active blackbody temperature reg ulation possible used as a part of a typical variable target test system based on off axis collimator 2 Thermal imager is looking into a big blackbody located within focus range of the tested imag
64. Fig 4 12 Multi USAF 1951 target recommended for measurement of resolution of collim ators Experiments carried out by the author of this book with infrared collimators used in different test systems showed that the center spatial resolution of these col limators varied significantly from about 30 Ip mrad to over 300 Ip mrad The reas ons for this significant dispersion of resolution of infrared collimators are different aging processes manufacturing errors of the mirrors alignment errors deteriora tion of coating properties dust on mirror surfaces etc Now let us check if such collimators can be used as important blocks of measuring systems for testing thermal imagers We can easily calculate the minimal acceptable collimator resolu tion using formula 4 2 and some basic data of several thermal imagers The calculation results are shown in Table 4 4 We can make two basic conclu sions from the data presented in this table First the requirements on spatial resolution of IR collimators significantly depend on a field of view of the tested thermal imagers The requirements are very low in case of imagers working in wide field of view mode but they are many times high er in case of the same imagers working in narrow field of view mode Second in case of testing short medium range thermal imagers the collimators of resolution at 25 Ip mrad can be considered as acceptable Collimators of resolu tion about 50 Ip mrad enable to carry out
65. However the safest way to evaluate surveillance thermal cameras is to measure MRTD of this thermal im ager and calculate the detection recognition and identification ranges of a standard NATO target using methodology proposed by this standard STANAG 4347 Defin ition of nominal static range performance for thermal imaging systems 1995 The standard defines precisely parameters of the standard target standard atmo spheric conditions and presents a way to calculate the detection recognition and identification ranges of the standard target on the basis of the MRTD function 36 of the tested thermal camera Two well known computer models that enable calcu lations of performance ranges for thermal imagers NVTherm or TRM3 use al gorithms that can be treated as modified versions of the method proposed by the STANAG 4347 The detection recognition and identification ranges of the standard NATO tar get are potentially a good criterion to be used in requirements on surveillance thermal imagers It is apparently a good idea to present requirements on a surveil lance thermal imager by presenting requirements on the performance ranges Table 3 2 However in order to calculate the performance ranges of a thermal imager we must know its MRTD function Therefore a more common way to specify re quirements on surveillance thermal cameras is to present requirements on MRTD characteristic in a form shown in Table 3 3 Table 3 2 Exemplary requireme
66. MRTD using a variable distance test system but keep ing the same measurement method used by variable target test systems This means by keeping constant spatial frequency of the target and changing target temperat ure difference during measurement of a single MRTD point It is however a rather slow measurement procedure When we change temperature difference of the black body we must wait typically about 60s for temperature difference to stabilize If we consider that we need often ten steps or more to find proper temperature difference then we see that measurement of a single MRTD point can take a dozen or more minutes Therefore in case of variable distance systems MRTD is typically measured by keeping temperature difference constant but changing spatial fre quency distance even for a single MRTD point If we can change the distance quickly then we can save a lot of time by using this measurement method There fore this method is highly recommended in case of testing small portable thermal imagers that can be easily moved In case of testing bigger thermal imagers it is re commended to regulate a distance by changing position of the test system placed on a movable platform It is preferable if the test system is battery powered Due to the presented above advantages of the constant temperature variable spa tial frequency method only for this method will be presented Next the measure ment are shall be only for positive contrast targets Fur
67. Michael C Dudzik ed SPIE 1993 3 Holst G C Testing and evaluation of infrared imaging systems JCD Publishing Company 1993 4 STANAG 4349 Measurement of minimum resolvable thermal difference MRTD of thermal cameras 1995 5 www cl systems com 6 Www eoi com 7 www hgh fr 8 www inframet com 9 www optikos com 10 www sbir com 156 6 STANAG 4349 requirements STANAG 4349 present some technical requirements on equipment used for testing thermal imagers Here there is presented a table with a list of these requirements and comments about technical capabilities of modern test systems The comments can be useful for laboratories that implemented quality systems according to ISO EN standards and which need to prove that their test system fulfill requirements of this well known standard Table 6 1 Comments on Stanag 4349 4 The MRTD is the minimum temperat All manufacturers of test equip ure difference which allows an observer ment use the same definition of to resolve a test pattern in accordance MRTD with a given criterion MRTD is a func tion of spatial frequency of the test pat tern NOTE 1 The MRTD measured at a given Test software corrects measure temperature can be converted to the ment results depending on ambi MRTD at other temperatures if neces ent temperature and spectral re sary The conversion factor depends on sponse of tested thermal imager the spectral response of the thermal im
68. TESTING THERMAL IMAGERS Practical guide Krzysztot Chrzanowski Military University of Technology Warsaw Poland TESTING THERMAL IMAGERS Practical guidebook Krzysztof CHRZANOWSKI Warsaw 2010 Copyright 2010 Krzysztof Chrzanowski All rights reserved No part of this book may be reproduced in any form by any means without permission in writing from the copyright owner Publisher Military University of Technology 00 908 Warsaw Poland INSB 978 83 61486 81 7 Printed by P P H Remigraf 11 Ratuszowa Str 03 430 Warsaw Poland Acknowledgement Research needed to obtain results presented in book was partially financed by the grant from the Polish Ministry of Science and Higher Education no 6 ZR8 2008C 07035 Notice The Publisher and Author have taken great care in preparing the information and guidelines presented in this book However the guidelines and other material within are of a general nature only The Publisher and Author take no responsibility with respect to the use of the information guidelines and material furnished and assume no responsibility for any damages or costs sustained while using the guidelines The Publisher and Author further disclaim any and all liability for any errors omissions or inaccuracies in the information guidelines and material given in this book whether attributable to inadvertence or otherwise and for any consequence arising there from Warsaw 2010 Author s Preface T
69. Thales An genieux As we can see in Table 2 1 the Gen 2 thermal cameras are characterized by sig nificantly better thermal and spatial resolution that the Gen 1 thermal cameras This means that quality of the image and sensitivity offered by the latter cameras is signi ficantly inferior However situation is not so clear if we compare Gen 2 and Gen 3 cooled thermal cameras Thermal sensitivity of Gen 3 cooled thermal cameras is usually at least slightly better that of thermal resolution of Gen 2 cameras How ever image resolution of modern Gen 2 thermal cameras is superior to image resol ution of typical Gen 3 cameras based on 320x240 FPA particularly in a horizontal direction This inferiority of Gen 3 cameras can be eliminated by the use of micro scanning technique that can improve up to two times image resolution in both ho rizontal and vertical directions However the disadvantage of microscanning tech nique is the higher production costs and reduced reliability The inferiority of im age quality offered by typical Gen 3 thermal cameras in comparison to Gen 2 cam eras can be fully eliminated if 640x512 or bigger FPAs are used A generation number is not connected strictly with image quality it is more con nected with mass dimensions manufacturing costs and reliability of the thermal camera The generation number suggest rather potential of the detector module but does not describe quality of a thermal camera Next in order to
70. The temporal noise refers to temporal variations of the signals generated by detector pixels during observation of an uniform target variations of the signal in a single line for case of scanning cameras or frame to frame variations of pixels signals for case of staring cameras The spatial noise refers to differences between the signals generated by different pixels during ob servation of an uniform target that does not change from frame to frame Both types of noise have its own noise power spectral density NPSD Noise is a complex phenomenon difficult for characterization In general we can find in literature three different noise analysis approaches to characterize the noise phenomenon present in thermal images 1 Three dimensional noise model where the noise is divided into eight com ponents Visualization is in form of multi dimensional images or numbers 2 Noise phenomenon is characterized by a single parameter presented as a number 44 3 Noise phenomenon is characterized by four parameters presented in form of numbers 3 3 1 3D noise model The 3D noise model is based on the concept of the D directional averaging op erators that allow the mathematical derivation of eight noise components from the noise data set 8 The operators average the data in the direction indicated by the subscripts Let us assume that a sequence of images generated by the tested imager was captured Then the captured data can be presented in form
71. a noise phenomenon is too soph 48 isticated to be commonly accepted and used It seems that in this situation a middle way when the noise is characterized by a set of three four parameters is an optimal solution Four parameters approach is based on the assumption that the noise present in images generated by thermal cameras can be generally divided into two groups temporal noise and spatial noise Next each group can be further divided into low frequency noise and high frequency component Noise ES Y Temporal noise Spatial noise Low frequency High frequency Low frequency High frequency temporal noise temporal noise spatial noise spatial noise 1 f noise NETD non uniformity FPN Fig 3 20 Types of noise Temporal noise generates temporal variation of intensity of camera pixels even when target radiation does not change in time Spatial noise generates spatial variations of intensity of camera pixels even in case of uniform target filling the camera field of view Low frequency temporal noise generates slow temporal variations of intensity of camera pixels This noise component creates an effect called 1 f noise The latter component is noticeable if we capture and compare the images generated by the camera separated by a relatively long period of time say at least a dozen or more minutes We can see in Fig 3 21 that some groups
72. aboratory conditions in a situation when they are expected to work properly at extreme ambient temperatures It is a com monly forgotten truth that parameters of thermal imagers can vary significantly with ambient temperature This dependence or in other words imager temperature stabil ity can be measured by placing the tested imager inside a temperature chamber of variable ambient temperature and then by measuring imager characteristics There are two methods to measure temperature stability the first by putting to the chamber both the imager and the blackbody the second by putting to the chamber the imager and using a chamber with a transparent window The latter method is more convenient for the test crew There are many commercially available temperature chambers on the market However they are typically not optimized for testing thermal imagers they are too big have too high thermal temporal inertia and the chambers often are not equipped with infrared transparent windows Therefore it is recommended to use for testing thermal imagers small temperature chambers of low thermal inertia equipped with an infrared transparent window Temperature range of the temperat ure chamber should fit to the temperature range of the environment where the tested imager is to be used Such a temperature chamber can be also very useful during calibration process spatial noise correction during manufacturing process 4 2 6 2 Optical table Special expen
73. about 50 mm Next blackbody for the test sys tems should simulate objects of rather low temperature to simulate typical observa 92 tion condition Further on both positive and negative temperature differences between the target and the blackbody are needed during testing thermal imagers Because of these requirements thermoelectric area blackbodies dominate among blackbodies used in systems for testing thermal imagers Thermoelectric area blackbody is an blackbody that uses Peltier element for tem perature control of the emitting element Peltier element Peltier module is a semi conductor based electronic component that functions as a small heat pump By ap plying a low voltage DC power source to a Peltier element called also often TEC thermoelectric cooler because of its ability to cool heat will be moved through the element from one side to the other One face of the element will be cooled while the opposite face simultaneously is heated Consequently a thermo electric element may be used for both heating and cooling by reversing the polarity changing the direction of the applied current This ability makes TECs highly suit able for precise temperature control applications as well as where space limitations and reliability are paramount Due to ability to heat or cool relative to ambient tem perature the thermoelectric blackbodies are usually called differential blackbodies There are a few important advantages of differential bl
74. ackbodies that made this type of blackbodies an ideal choice for testing modern thermal imagers First standard temperature range of differential blackbodies from about 25 C to about 75 C overlaps the temperature range needed during tests of surveillance thermal imagers Second it is possible by careful design to develop differential black bodies of excellent temperature resolution 0 1 mK stability 1 mK and uniformity 10 mK Third these blackbodies are characterized by low mass and low temporal inertia in contrast to cavity blackbodies for the same temperature range typically li quid based blackbodies or pipeline blackbodies All these features created situation when differential thermoelectric blackbodies are practically the only type of black bodies used by manufacturers of professional test systems for testing surveillance thermal imagers Please note however that the cavity blackbodies are frequently used in testing accuracy of commercial thermal imagers to be used for non contact temper ature measurements due to wider temperature range at high temperatures 4 2 2 1 Design Differential blackbodies to be used as components of test systems are offered by different manufacturers 24 26 30 32 29 Now we will discuss in details design of a differential blackbody offered by one of manufacturers 30 but the conclusions will be generally valid for all high performance differential blackbodies present on the market The blackbodies
75. acquisition analysis module is built from the following blocks PC frame grabber video card and test software It is practically a specialized PC to carry out tasks needed in testing thermal imagers The module should enable acquisition of the output signal from the tested thermal imager analysis of the captured images and semi automatic determination of important characteristics of thermal imagers 4 2 5 1 PC In general PC should enable processing of the input data from the frame grab ber and calculation and visualization of characteristics of thermal imagers Practic ally all modern PCs can handle such tasks 106 4 2 5 2 Frame grabber Almost all surveillance thermal imagers generate output images in form of ana log video electrical signal PAL standard or NTSC standard However more and more thermal imagers generate output images in form of digital signal Fire Wire USB 2 0 Camera Link GigE LVDS Especially the latter three standards Camera Link GigE LVDS are popular for designers of high resolution high speed thermal imagers The task of a frame grabber video card is to capture sequences of images gen erated by tested thermal imager The ideal frame grabber should accept input data in all typical electronic standards PAL NTSC Fire Wire USB 2 0 Camera Link GigE LVDS Next it is critical that there should be no noticeable degradation of image quality caused by the frame grabber Please note that most low cost fr
76. adiator equipped with fans The task of the cooling radiator is to dissipative quickly the heat from the Peltier element Measurement of temperature of the emitter plate and temperature of the front wall is done by measurement of temperature dependent resistance of the PRT resistors The resistance is converted to voltage using a resistance bridge made from resistors of high linearity temporal stability and temperature stability Next the output signal is amplified using a low noise preamplifier with corrected temperature drift Finally the analogue voltage is converted to digital signal using 24 bit A D converter and that digital signal is sent to the controller The controller block consists of three basic modules micro controller no 1 mi cro controller no 2 and power supply module RS 232 Peltier termoelement RTCB gt microcontroller 1 microcontroller 2 screen keyboard power supply module Fig 4 15 Block diagram of the controller block There are 3 basic functions of the micro controller no 1 First it enables com munication with the radiator block with PC computer through RS 232 port and with the micro controller no 2 Second the micro controller no 1 converts val ues of the output voltage into values of temperature Third the micro controller no controls through the D A converter the voltage applied to the Peltier ther moelement in the radiator block The micro controll
77. ael C Dudzik ed SPIE 1993 Holst G C Testing and evaluation of infrared imaging systems JCD Publishing Company 1993 International Lighting Vocabulary CIE Publ No 1 7 4 IEC Publ No 50 845 1987 International Vocabulary of Basic and General Terms in Metrology International Organisation for Standarisation 1993 ISO 15529 Principles of measurement of modulation transfer function MTF of sampled imaging systems 1999 Miller J L An introduction to infrared cameras 1999 The photonics design and applications handbook Laurin Publ Co p 133 139 2000 Pruchnic et al Design analysis and testing of collimators used for testing low resolution thermal imagers Proc SPIE Vol 2743 p 337 348 Infrared Imaging Systems Design Ana lysis Modeling and Testing VII Gerald C Holst Ed Publication Date 6 1996 Qin Y et al Novel quantitative NDT method for composite structures Proc SPIE Vol 3173 p 274 278 Ultrahigh and High Speed Photography and Image based Motion Measurement Donald R Snyder Andrew Davidhazy Takeharu Etoh C Bruce Johnson James S Walton Eds Publication Date 12 1997 Rodgers A L Surveillance and target acquisition systems Brassey s Battlefield Weapons Sys tems amp Technology Vol VI Brassey s Publishers Ltd 1983 Fourier Ron Accuracies in FLIR test equipment 7th Mtg in Israel on Optical Engineering SPIE Vol 1142 pp 109 117 1991 16 23 24 25 26 27 28
78. age They can be treated as optical intensifiers of a scenery at low illumination conditions to the levels when human eye can carry out effective surveillance TV cameras are electronic devices that convert the input image into an electrical signal that after electrical amplification generate the output image in analog or di gital formats used by television or video technology The TV camera technology offers inherent capabilities to record process and transmit image of the observed scenery There are many types of TV cameras color nono CCD cameras color mono CMOS cameras intensified charge couple device ICCD cameras cooled CCD cameras and electron bombarded charge couple device EB CCD cameras silicon intensified target SIT tube cameras intensified silicon intensified target ISIT tube cameras etc Thermal imagers are the imaging systems sensitive to mid wave and long wave infrared radiation that generate images of the observed scenery using thermal radi ation emitted by the scenery The origin of all three groups of electro optical systems is connected with milit ary applications However at present these imaging systems are used in high num bers of military and civilian sectors According to a type of radiation used to create an image of the observed scenery electro optical imaging systems can be divided into two distinct groups the sys tems that create image using the radiation emitted by the observed targets and the syste
79. al con ductivity mostly metal alloys and regulate temperature of this block using a heat ing cooling element Emissivity of cavity blackbodies is typically over 0 995 even when emissivity of the cavity walls is much lower about 0 5 0 8 Most medium or high temperature blackbodies are designed using the cavity concept Typical commercially available cavity blackbodies are characterized by relatively small emitting aperture about 1 inch aperture and quite high temporal inertia Second method is to put a layer of high emissivity material over a flat metal ele ment of regulated temperature Some special black paints of emissivity equal to 0 97 are put over blackbody emitters The latter elements are manufactured from high conductivity metals usually copper It is possible to use the second method to achieve bigger emitting apertures then using the first method The blackbodies designed using the second methods are called area blackbodies The area blackbod ies are also characterized by lower temporal inertia than the cavity blackbodies due to smaller mass However due to lack of high emissivity coatings resistible to high temperatures the area blackbodies can be used only at low temperatures typically the maximal temperature is no more than 400 C The emitting area of the blackbodies used in systems for testing thermal imagers must be bigger than the target plates This creates requirement that emitting area must be typically at least equal to
80. ame grabbers are designed with the aim to be used in applications where quality of the captured images is not critical but compression ratio is the most important paramet er Each frame grabber device should provide a proper software driver which should contain routines compatible with one of existing commonly used APIs Application Programming Interfaces Image acquisition applications software modules are based on such standards as e g TWAIN or more native to Microsoft s operating systems interfaces as DirectShow or WIA Windows Image Acquisition Main functionality of image acquisition software module is to get image from video capture device in form of separate frames or in most cases in form of video sequence Such collection of image data can be passed on for further processing and analysis 4 2 5 3 Test software There are four tasks of test software used in systems for testing thermal cameras 1 Remote control from PC of the test system hardware blackbody rotary wheel 2 Acquisition of the output images generated by the tested thermal camera 3 Software support during measurement of subjective parameters MRTD MDTD 4 Semi automatic measurement of objective parameters of tested thermal imager The tasks mentioned above can be handled in different way by different computer programs Here we will formulate basic requirements and recommendations for test software 1 Easy to learn graphical method for control of b
81. andardized in the International Lighting Vocabu lary published by the International Lighting Commission CIE and the International Electrotechnical Commission CIE in 1987 15 However there are vast areas of the infrared technology where terminology is not standardized mostly due to the fact that scientists and engineers of completely different background work nowadays in infrared technology It results in situation when different authors use different ter minology in scientific papers manuals and catalogs making them difficult to under stand even for professionals Such a situation is particularly difficult for newcomers to this technology and non native English speakers Some examples will be dis cussed next Firstly we will start with the term electro optical system used quite often in this chapter This popular mostly in USA and Asia term is used in this book but the terms optoelectronic systems or optronic systems have the same mean ing and are quite often used in literature too Next the term electro optical sys tem in general refers to any system that uses optical radiation infrared radiation 9 visible radiation and ultraviolet radiation We used this convention and classified thermal cameras night vision devices TV cameras as electro optical systems However it must be remembered that in many literature sources electro optical systems are understood as only night vision devices technology or technol
82. aperture of application non optics of the tested convenient in case of imager high cost expanded tests On the basis of earlier presented analysis of features of three different types of test systems the recommendations on optimal application are presented in Table 4 2 79 Table 4 2 Recommended application area of different types of test systems Recommended applications Not recommended applications Variable test system expanded testing of tests at field conditions surveillance thermal imagers at laboratory conditions Variable distance test Simplified tests of portable expanded tests of systems thermal cameras at indoor surveillance thermal conditions using long imagers corridors Simplified tests at field conditions Variable target distance test tests of measurement tests of long range system commercial thermal surveillance imagers imagers 4 2 Blocks of test system First thermal imagers were designed for military applications to enable obser vation of long distance targets Minimal focus distance of these imagers was often more than 100 m Therefore testing of such imagers was possible only using vari able target test systems that used a collimator as an image projector Now due to presence on the market of a large number of thermal imagers designed also for short distance surveillance or for non contact temperature measurement it is often pos sible to use
83. asily fulfilled if air condi tioning is used Next please note that the standard allows to carry out measurement at ambient tem peratures outside this range and to correct the measurement results Transmittance of the collimator emissivities of blackbody emissivity of the 4 bar targets are known and they are taken into ac count by test software when cor rected MRTD results are calcu lated 5 4 In carrying out the measurements the observer may optimise results by altering both the illumination level in the room and any settings on the imager adjusting the distance of his eye from the display and slightly altering the position of the test pattern within the field of view If a monitor is used it must be of the type normally used with that imager and it is recommended that once set up it should not be adjusted during measurement 6 1 Starting from an invisible test pat tern the method is based on the determin ation of the temperature difference or required to make the test pattern visible Owing to limitations in the test equipment the indicated temperature dif ferences may not be the true temperature differences between test pattern and black body Therefore the following procedure is recommended to establish the MRTD starting with the temperatures of the test pattern and black body approximately equal see Note 2 the temperature of the black body shall increase until the test pattern appears in positiv
84. asure the classical subjective MRTD However because determination of K v requires testing of many thermal cameras at least 20 30 then the objective MRTDatw can be determined only in 3 2 33 case of large production lines In case of typical tests of small quantities of different thermal imagers only classical subjective MRTD can be measured Nowadays classical MRTD is considered as the most important parameter of thermal imagers and MRTD is typically used for range predictions for real targets However it was reported many times that the MRTD concept when applied to undersampled imagers generates incorrect range predictions particularly detection range The biggest problem is low accuracy of performance modeling over Nyquist frequency There are at least three competing solutions to eliminate the mentioned above limitation of MRTD concept and to improve accuracy of range prediction The DMRT Dynamic MRTD is an MRTD variation method that assumes that the bar pattern is moved relative to the sampling lattice with optimum speed about 4 pixel per frame This measurement way enable for human eye to integrate over various phase positions of the 4 bar target and the the target is perceived at spatial frequencies significantly higher than Nyquist frequency However this artificial movement of the target during measurement creates conditions that are drastically different than conditions during static observation Dependence of the test pr
85. ater in details However it should be reminded that the edge method is a modified slit method because differentiated Edge Spread Function equals Line Spread Function Finally we should remember that if properly applied then both two measurement methods should generate the same MTF measurement results Typical simplified measurement principle used by most modern test systems for MTF measurement is presented below 1 Projection of an image of the edge target by the collimator into direction of tested imager 2 Tested imager generates a blurred copy of projected image of the edge tar get 3 Frame grabber captures images generated by the tested imager 4 Test software using the frame grabber captures images of the edge target generated by the tested imager 5 Test software carries analysis of the captured images calculates MTF function and presents the measured data in a graphical form or in form of a table Practical implementation of this algorithm is not so simple as there are high re quirements on both collimator target blackbody frame grabber and test software that must be fulfilled to generate accurate measurement results e Collimator to generate image of the edge target without any noticeable de gradation of image quality e Blackbody to generate a stable uniform image of the edge target require ment on temperature temporal stability and temperature uniformity e Edge target the angular differences between the real e
86. ating are similar like in case of the tar gets Requirements on positioning accuracy are not high in case of typical tests of thermal imagers Slight angular variations of position of the image projected by the collimator has no influence on measurement results of parameters of thermal im agers It is enough if positioning repeatability of the wheel is not worse than about 1 mm However there are some tests of thermal imagers used for automatic target recognition tasks when much better positioning repeatability close to 0 1 mm level is required Fig 4 17 Photo of a motorized rotary wheel and differential blackbody courtesy of In framet During tests of thermal imagers we need to know temperature of the blackbody radiator and temperature of the target Temperature of the blackbody radiator is typ ically measured using a small temperature sensor inserted to a hole in the radiator This direct contact measurement method cannot be used to measure temperature of the targets because the targets are rotating One solution to solve the problem is to use a sliding temperature sensor that slightly touches the active target and measures its temperature However that direct contact of the sensor with the target when the latter is moving generates some heat and this effect creates additional measurement error Another solution is to measure target temperature indirectly The sensor is at tached to the rotary wheel If there is good thermal
87. avities is created in the surface of the emitter Two high quality platinum resistance thermometers PRT resistors are placed in a hole drilled in the emitter plate and in the hole drilled in the front wall of the of the RTCB radiator The task of the second sensor is to measure the ambient temper ature The PRT resistors are characterized by high linearity temporal stability and the high temperature coefficient Due to good thermal contact of the PRT resistor with the emitter plate or with the front wall it is possible to measure accur ately temperature of the emitter plate and temperature of the front wall or further of the rotary wheel when the wheel is connected to the radiator block Because of temporal inertia effect the measurement is typically done with frequency not high er than 1 Hz Next due to fixing of the PRT resistors in high thermal conductivity 94 material the influence of the effect of resistor self heating on its resistance is min imized Further on the emitter was placed inside a cavity made from low thermal conductivity material in order to minimize influence of air random vortexes on temperature distribution on the surface of the emitter The back side of the emitter plate is closely attached to the Peltier thermo ele ment of temperature dependent on the applied voltage It is possible to heat or cool the emitter applying proper voltage The other side of the Peltier thermoelement is closely attached to the cooling r
88. bout Author Krzysztof Chrzanowski received his Ph D and D Sc both in Electronics from Military University of Technology in Warsaw Poland He works currently as Pro fessor in the mentioned above university He is also CEO of a spin off company Inframet www inframet com His main scientific interests include analysis test ing and computer simulation of electro optical surveillance systems thermal im agers night vision devices TV cameras laser range finders multi sensor surveil lance systems non contact thermometry and general metrology He is an author or co author of over 100 scientific papers and conference communications This books presents knowledge of the author on testing thermal imagers that was accumulated during over two decades of scientific work in the field of electro op tical technology interconnected with a series of practical projects The author hopes that this book can become a practical guide in field of testing thermal imagers for a wide community of people interested in this fascinating technology of thermal imaging 164 Krzysztof Chrzanowski received his Ph D and D Sc both in Electronics from the Military University of Techno logy in Warsaw Poland He works currently as Professor in the mentioned above university He is also CEO of a spin off company Inframet www intramet com His main scientific interests include analysis testing and computer simulation of electro optical surveillance systems
89. c is increasing quickly Thermal imagers are often pur chased within big tenders of total value over one million EUR There are sometimes doubts whether the offered or delivered thermal imager fulfill technical specifica tions of the tender Results of testing thermal imagers offered in such tenders are very important for decision makers Secondly a number of thermal imagers used in automotive industry increases quickly It is still the beginning of a long process but even now the automotive in 13 dustry is one of the most important application areas of thermal imagers The main criterion for the automotive thermal imager is its cost but image quality is still im portant and proper testing is needed Thirdly a price of infrared focal plane areas IR FPA decreased significantly during the last decade It is particularly true in case of non cooled IR FPAs Due to this reason the number of manufacturers of thermal imagers has increased signific antly The only ways for these new manufacturers to be sure about real quality of their products is to test them and compare test results with the parameters of thermal imagers offered by their competitors Fourthly the competition on the market of thermal imagers is strong Good mar ket position can be reached and kept only if an imager of good ratio quality to price is delivered In order to keep stable and known quality the extensive and accurate testing during production line is needed Further o
90. calculate the detection and recognition ranges on the basis of the measured and known MRTD function of the tested thermal camera using the recommendation from STANAG 4347 The summary of the recommend ations is shown in Table 3 4 Table 3 4 Target parameters atmosphere conditions and resolution criteria specified in the standard STANAG 4347 Target Atmosphere Resolution criteria according to 50 probability size Rectangle transmis T R e oR detection 1 line 2 3x2 3 m sion law Redisanceii pair tar kii get o coefficient temper AT 2 K re o at good 0 2 km recogni 3 line ature dif lated to black atmospher tion pair tar ference body temperat _ ic condition get ure of 288 K o atlim 1km identify 6 line ited atmo cation pair tar spheric get condition In details the detection recognition and identification ranges of the standard NATO targets can be calculated using the below presented algorithm 1 Convert MRTD characteristic into a new one by changing variable from spatial frequency v mrad to the range R km using the following for mulas arden km 2 3 v mrad re km 2 3 3 v mrad Ria km 2 3 6 v mrad 2 Calculate decrease in the initial temperature difference AT it was assumed that initial AT 2 K between the target and the background due to lim ited atmospheric transmission AT R ATye 3 3 38 3 Determine the respec
91. cameras build using 2D ar ray detectors cooled FPA based on InSb HgCdTe QWIP technology or non cooled FPAs based on microbolometer or pyroelectric ferroelectric technology that have at least 10 element on the focal plane These staring arrays are scanned elec 21 tronically by circuits integrated with the arrays These readout integrated circuits ROICs include e g pixel deselecting antiblooming on each pixel subframe ima ging output preamplifiers and some other functions The opto mechanical scanner is eliminated and the only task of the optics is to focus the IR image onto the mat rix of sensitive elements Fig 2 3 Exemplary Gen 2 thermal camera Sophie courtesy of Thales Optronique 2002 0280 Fig 2 4 Exemplary cooled Gen 3 thermal camera Catherine XP courtesy of Thales Optronique Third generation thermal cameras have been offered since the beginning of the 90s to compete with their predecessor First they have been offered as cooled MWIR cameras using InSb or HgCdTe technology sensitive in 3 5 um atmo spheric window in situation when for most geographic conditions LWIR thermal cameras are desirable Cooled LW IR Gen 3 thermal cameras based on QWIP tech nology started to be commercially available at the end of the 1990s Almost at the same time non cooled thermal cameras based on microbolometer and pyro electric ferroelectric technologies became fully commercially available Image 22 quality of n
92. ces can be distorted or even damaged when subjected to wide temperature changes Therefore low thermal expansion substrate materials are critical to suc cessful performance of imaging reflective optical systems to be used at different ambient temperatures Low expansion borosilicate glass LEBG known by the Coming brand name Pyrex is well suited for high quality front surface mirrors designed for low optical deformation under thermal shock Pyrex coefficient of thermal expansion is lower than in case of optical crown glass Synthetic fused silica has a very low coefficient of thermal expansion Fused silica mirrors can be polished to extreme accuracies thereby minimizing wavefront distortion and scattering Zerodur is a unique glass ceramic material whose thermal expansion is almost zero This stability is essential in diffraction limited systems where the optical fig ure must not vary under thermal changes Parameters of the described earlier materials used for mirrors substrates are shown in Table 4 5 As we can see the best material for the substrate of mirrors seems to be Zerodur due to negligible thermal expansion and ultra high surface ac curacy However Zerodur is also the most expensive material of the four analyzed materials Surface mirrors accuracy at the level of over A 12 is needed only in rare applications of testing optics for space applications In case of collimators for test ing thermal imagers surface accuracy at the level
93. chieved by turning the eyes towards the current point of interest in the field of view Each cone cell is built using three color sensitive pigments red sensitive pig ment green sensitive pigment and blue sensitive pigment In the diagram above the wavelengths of three types of cones red green and blue are shown The peak absorption of blue sensitive pigment is about 445 nm for green sensitive pigment about 535 nm and for red sensitive pigment about 570 nm The pigments generate three electrical signals a B and y A brain analyzes the signals and determines the color of the incoming light on the basis of the ratio between a B and y signals Next the perceived brightness depends on the sum of the signals a P and y When the signals a B and y are almost the same then the target will be perceived as white gray or black white high sum of the signals gray medium sum of the sig nal black very low sum of the signals The human eye can sense many gradation of color when red green and blue signal are mixed It is usually considered that hu man can sense about 380 000 color gradations Fig 1 3 Principle of color vision Power of the light coming to the light detectors distributed on the retina surface is automatically regulated by the iris This component can change its diameter from about 2 mm to about 10 mm and this phenomenon enables us to work at different il lumination conditions The distribution of the con
94. components resolution imaging spatial resolu tion NPSD noise power spectral density subjective parameters based on resolution tar gets 30 No Accuracy para Geometric para Operational Spectral parameters meters meters parameters 1 accuracy Field of view Focus range Spectral sensitivity func Minimal Error tion 2 NETD Noise Magnification Eye distance Equivalent Tem perature Differ ence Noise Gen erated Error 3 Temperature sta Distortion Diopter settings bility 4 Slit Response Image rotation Function 5 Boresight align ment Attention there is no correlations between parameters from different groups having the same number Numbering was used only to make easier identification of different parameters Accuracy parameters are necessary to evaluate commercial thermal cameras to be used as non contact thermometers They are useless in case of thermal imagers to be used in surveillance applications Measurement of operational parameters of thermal cameras does not differ sig nificantly from a measurement of the same parameters of other visual imaging sys tems There is the same situation in case of spectral parameters Therefore both op erational parameters and spectral parameters will not be discussed here 3 1 Subjective image quality parameters Thermal cameras are imaging systems used to enhance human ability to see
95. d 1 mrad f 1000 mm f 1500 mm f 2000 mm f 3000 mm 4 00 0 13 0 19 0 25 0 38 3 36 0 15 0 22 0 30 0 45 2 83 0 18 0 27 0 35 0 53 2 38 0 21 0 32 0 42 0 63 2 00 0 25 1500 0 50 0 75 1 68 0 30 0 45 0 60 0 89 1 41 0 35 0 53 0 71 1 06 1 19 0 42 0 63 0 84 1 26 1 00 0 50 0 75 1 00 1 50 0 84 0 60 1500 1 19 1 79 0 71 0 70 1 06 1 41 2 11 0 59 0 85 1 27 1 69 2 54 0 50 1 00 1 50 2 00 3 00 0 42 1 19 1 79 2 38 3 57 0 35 1 43 1500 2 86 4 29 0 30 1 67 2 50 3 33 5 00 0 25 2 00 3 00 4 00 6 00 0 21 2 38 3 57 4 76 7 14 0 18 2 78 4 17 5 56 8 33 0 15 3 33 1500 6 67 10 00 0 13 3 85 5 77 7 69 11 54 0 10 5 00 7 50 10 00 15 00 122 Table 5 3 Spatial frequencies in lp mrad of two 4 bar patterns as a function of the distance R between the imager and the variable distance test system and the width of target bar a Rim a 10 a 4 R a 10 a 4 R a 10 a 4 R a 10 a 4 mm mm m mm mm m mm mm m mm mm 1 10 05 0 125 19 10 95 2 375 37 1 85 4 625 60 3 7 5 2 0 1 0 25 20 1 2 5 38 11 9 4 75 62 13 1 7 75 3 10 15 0 375 21 1 05 2 625 39 1 95 4 875 64 13 2 8 4 10 2 0 5 22 1 1 2 75 40 2 5 66 3 3 8 25 5 10 25 0 625 23 1 15 2 875 41 2 05 5 125 68 13 4 8 5 6 0 3 0 75 24 1 2 33 42 2 1 5 25 70 13 5 8 75 7 10 35 0 875 25 1 25 3 125 43 2 15 5 375 72 3 6 9 8 10 4 1 26 1 3 3 25 44 1
96. d the effective resolution ER is defined as 1 2 N The effective resolution is presented using angle units in case of thermal cameras typically millionaires are used Third experiments have shown that perceived image sharpness is closely re lated to the spatial frequency where MTF is 0 5 This means that the spatial fre quency at which MTF drops to 0 5 can be a good indicator of imager quality This frequency is called the half frequency HF It is expressed in spatial frequency units Fourth the effective instantaneous field of view EIFOV is defined as ovi 3 20 2 HF EIFOV is presented in angle units typically in millionaires Fifth the limiting resolution is defined as spatial frequency at which MTF equals from about 0 02 to 0 05 The definition is based on the fact that humans usu ally cannot distinguish high contrast sine pattern at frequencies where MTF drops below the level 0 02 0 05 Exact value of the limiting MTF level depends on the observer As we can see determination of these five numerical parameters related to MTF function is quite easy when the latter function is known However nowadays in era ER 3 19 60 of computer technology when it is easy to measure present and store MTF function it is better always to use MTF function as original data The numerical parameters listed earlier should be used only for comparisons of thermal cameras of different MTF functions when we need to have a simpl
97. d above data 1 Thermal imager fulfills requirement only in case A This is the worst scenario for manufacturer but very safe situation for the final user 2 Thermal imager fulfills requirement in case A B and D This scenario assures that no good imager is classified as non fulfilling specifications but this scenario is very risky for the final user 3 Thermal imager fulfills requirement in case A B Now the risk due to limited accuracy of MRTD measurement method is divided fifty to fifty between the manufacturer and the final user The decision which interpretation method should be applied to the MRTD test results should be determined immediately when MRTD requirements are decided As we see in the example presented earlier not only MRTD measurement results are important but also method of their interpretation 5 2 MTF 5 2 1 Measurement concept Modulation Transfer Function MTF is a measure of sharpness or blurring of images generated by thermal imagers or in general by any imaging system There are over a dozen possible methods to measure MTF of imaging systems However in case of testing thermal imagers only two measurement methods are typically used Method one an image of a slit target generated by tested imager is acquired analysed and MTF of the tested imagers is calculated as a module from Fourier transform of Line Spread Function normalized blur distribution in the image of the narrow slit target MTF v F LSF x
98. d brain enables humans the ability to sense visible radiation light and to see The eye converts optical signals into electrical signals that are later sent by the optic nerves to the brain and finally after some kind of image processing we see the targets that are in our field of view The biological eye consists of two sys tems an optical system and a detection system The optical system cornea crystal line lens vitreous refracts and focuses the incoming light into the detection system retina The latter system contains millions of rods and cones which convert light energy into electrical signals sent to the brain via the optic nerve The retina con tains about 100 millions of rods and about 6 millions of cones The rods of relat ively low sensitivity are to work in daytime conditions the photopic vision and can deliver high resolution color vision The cones are high sensitivity detectors that work in night time conditions scotopic vision but they can deliver only lower res olution monochromatic vision lens array of detectors brain ae racessor Ne s z nerves P m gt D sa fibers Fig 1 2 Simplified optical diagram of human eye Human eye generates sharp image within rather narrow field of view about 10 because in the middle of the retina there is the highest concentration of rods and cones The rest of the field of view is rather blurry The experience of wide sharp human vision is a
99. d on subjective human ability to resolve some patterns 3 4 1 Parameters based on specifications of the IR FPA Infrared focal plane arrays are hearts of thermal imagers Therefore it is not strange that some IR FPA parameters like number of detectors number of lines detector angular size parameters called IFOV or DAS are often used to describe performance of thermal imagers 54 The first parameter is the total number of detectors of a two dimensional IR FPA sensor and it is used to describe resolution of staring thermal cameras The second parameter is the total number of vertical lines of a linear IR FPA This parameter is used to describe resolution of scanning thermal cameras IFOV instantaneous field of view or DAS detector angular substance in spite of different names are the same parameter defined as angular dimension of a single pixel of the IR FPA used in the thermal camera DAS mrad a um f mm 3 12 where a is the pixel linear dimension and f is the focal length of the optics Please note however that a pixel horizontal dimension can differ from a pixel vertical di mension Therefore there can be a difference between a horizontal resolution and a vertical resolution of thermal cameras It is very easy to get these simple data about the IR FPA sensor and the optics used in athermal imager Therefore not only manufacturers but also most community involved in infrared technology like to express imager resolution u
100. dard are presented The comments can be useful for laboratories that imple mented quality systems according to ISO EN standards and need to prove that their test system fulfill requirements of this well known standard Finally in Chapter 7 short guidelines for buyers of equipment for testing thermal imagers are given The guidelines can be useful for scientific manufacturing centers that analyze possibility of purchase of such test equipment 1 8 References 1 Anderson D M1A2 tank commander s independent thermal viewer optics optics design per spective Proc SPIE Vol 1970 p 128 138 Systems Oriented Optical Design Thomas U Kampe Joseph H Oberheuser Eds Publication Date 8 1993 15 9 10 11 12 13 14 15 16 17 18 19 20 21 22 ASTM standard E 1213 2002 Standard Test Method for Minimum Resolvable Temperature Difference for Thermal Imaging Systems ASTM standard E 1311 99 Standard Test Method for Minimum Detectable Temperature Dif ference for Thermal Imaging Systems Bryant Paul Grigor Jack McHugh Stephen White Steve Performance comparison of reflect ive and emissive target projector systems for high performance IR sensors Infrared Imaging Systems Design Analysis Modeling and Testing XIV SPIE Vol 5076 p 242 249 2003 Driggers Ronald Hodgkin Van A Vollmerhausen Richard O Shea Patrick Minimum resolv able temperature difference measurements
101. dge and the ideal edge must be many times smaller than the angular size of pixel of the tested thermal imager requirement on manufacturing accuracy of the target edge e Frame grabber to enable capturing image of the edge target generated by the tested imagers without any noticeable degradation of quality of the captured image requirement on ability to capture images from imagers 144 generating electronic image signals in different standards PAL NTSC FireWire USB 2 0 CameraLink GigE Test software user friendly software that enables calculation of MTF using advanced algorithms resistible to noise inherent in captured images As we can see typical test systems used for MRTD measurement built from the following modules collimator blackbody rotary wheel set of targets frame grab ber PC test software are also used for MTF measurements The only difference in hardware blocks is that the edge target is used for this measurement instead of typ ical 4 bar targets 5 2 2 Measurement procedure Different manufacturers use slightly different detail measurement procedures of MTF measurement Here we will present simplified typical MTF measurement procedure l Rotation of the rotary wheel with the targets using the control software un til the edge target is in the active position the target is at the collimator fo cal plane and can be seen by tested thermal imager The edge target should be vertical if horizontal MTF is to
102. digital filters Calculation method of four noise components from 3D patio temporal array depends on answer on a question what is dominant type of noise of thermal camera ergodic noise or non ergodic noise If the thermal camera is a source of ergodic noise then the detectors are considered as statistically dependent noise sources The same average ems will be measured if the average is calculated from n different detectors or the same ems noise of the same detector is the measured n times Then ems noise can be calculated as 2 2 2 Sit Sstt o js j5 2 a 3 6 ave n where s is the variance of noise from i detector or from the same detector but measured time n is the number of detectors or indicator how many times were re corded the data from the same detector If the thermal camera is non ergodic then the detectors are considered as inde pendent noise sources Each detector is statistically a different noise source Then ems noise is calculated as Set S t t S O S4 ave n 3 7 n Scanning thermal cameras can be considered to some degree as ergodic systems Staring thermal cameras are usually non ergodic systems In order to simplify 51 the analysis let us assume that the tested camera is non ergodic as staring thermal cameras dominate the market Anyway the consequences of non fulfilling this assumption are not really significant because the differences between ems noise values calculated
103. dsieusesnisesvatedestensedwassocsesesecednoshassensesace sosseaveaesdoosaned 68 3 7 Referentes cases cuskeiecscescesictnesaascaseacesonsesessestesesaeassescesaessaeesnsanaseusdebacsiancesestidessveteleases 70 4 Test equipment 5 seis isicacvavaceshesanateetes saacbasscadgecnisens ccecesancateesenceds 73 4 1 Types of test SySteMs cscsscssscsssscscesssssccsssssessssesessssscssssssessscesssessessssesssessseses 73 4 2 Blocks of test System ccccssccssssscssscsscssscsscssssssssssscssssssessssssessssssessessesssesssaees 80 4 22 Collimatorcsaicias dniietichs specie nit ea dill deeded de BAe needs 81 AD D Black BOd yrii e E E A a G 92 42 3 Rotary Wwheelanisnrsunneyiradinniiin io ea i a 99 42A VAT OGtS x ests e a e E A T E rhea tema eared 101 4 2 5 Image acquisition analysis module ssssesseseesesseseesessesesessrsssessseesseesserss 106 42 0 Optional DIGCKS a a r ene adie tei E oleh ae ee AT 109 4 3 Manufacturing R amp D support equipMent ccccccsessccsssssssscsssteceeecees 110 4 3 1 Correction of spatial MOISC ccccceeceesceeceesseeseeeeeseeseeeseceeeseseeeeseeeeeeteenseeaes 110 4 3 2 Calibration of measurement thermal imagelS c ccceeseesseeeeteeeeteeeeeneeeeens 112 4 3 3 Boresighting to a reference optical mechanical axis 113 AA RELCLENCES dsc scssscdesssescesentecoseocascendssvascasnscetnsonsessandessesestesseseaeecensesssnasesesesc iniadoeseage 114 5 Test pr ced reSi s is yiii sete
104. e contrast hot bars When the observer can just resolve the test pattern the temperature differ ence AT is recorded The black body temperature is then reduced causing the test pattern to disappear and then to re appear with negative contrast cold bars When the observer can just resolve the test pattern again the temperature differ ence AT is recorded see Note 3 The two measurements must be taken without delay so that temperature drift is avoided see Note 4 Note 2 The procedure can be speeded up by starting with a temperature difference between the test pattern and the black body of 0 8 AT both for positive and negative contrast 159 This is an requirement to be ful filled by users of test equipment not requirement on equipment This is an requirement to be ful filled by users of test equipment not requirement on equipment Offset is calculated in test soft ware that controls blackbody us ing the formula recommended in Note 3 Note 3 If it can be established that the temperature drift is negligible both posit ive and negative contrast measurement need to be made only for the lowest spa tial frequencies in order to determine the AT offset of the test equipment This AT offset is then calculated as the mean of the temperature differences for the positive and negative contrasts that is AT offset 0 5 AT AT and its value is then subtracted from the tem pera
105. e formulas for the coefficient cor only in the range from about 10 C to about 30 C Therefore it is recommended not to carry out MRTD outside this temperature range Influence of transmittance of the optical channel can be corrected by multiplying the results by a value of transmittance of the channel In case of a variable target test system the distance between the test system and the imager is very short influence of atmosphere on radiation propagation is negligible and transmittance of the optical channel is equal to collimator transmittance Values of collimator transmittance can vary quite significantly from about 0 85 to about 0 96 Next collimator transmittance depends on a spectral range Because of this significant variability the user should always get information from manufacturer about collimator transmittance in the spectral range of interest The correction is done using the formula presented below MRTDeor MRTD Toot 5 3 where MRTD is the corrected MRTD MRTD is the original measurement result and Tea is the transmittance of the collimator In case of variable distance test systems there is no collimator in the optical channel but the distance imager test system is longer In typical conditions when 119 the distance is below about 100 m and measurement is done at good atmospheric conditions we can consider that atmospheric transmittance is almost unity and does not carry out any corrections However when the measur
106. e measurement results is often difficult First there are no internationally accepted standards that could regulate meas urement on noise related parameters of thermal imagers There are different recom mendations in specialized literature Therefore there is a certain chaos in measure ment of noise parameters A prime example of this chaotic situation is a situation with the main noise parameter of thermal imagers the NETD According to one school NETD is a measure of only high frequency temporal noise and FPN is a measure of high frequency spatial noise However according to a second school NETD is a measure of both high frequency temporal noise and high frequency spa tial noise Thus it is necessary to be very careful in comparison of NETD para meter measured by two different test teams Second noise related parameters like NETD FPN NU depend on settings of tested thermal imagers mostly on Gain contrast and Level brightness but in most cases this relationship is not very strong However image processing like edge enhancement can change noise measurement results drastically By pressing the button Image enhancement on the imager keyboard the measured noise para meters can be increased even by a factor of two times Therefore it is recommen ded to compare several thermal imagers of the same type only if the tests were car ried out for the same settings of the tested imagers Third parameters of some IR FPAs used in m
107. e numerical criterion of comparison 3 4 3 Parameters based on imager response to point slit sources There are three numerical parameters of image resolution that are based on imager response to point slit sources 1 Point visibility factor PVF 2 Measurement spatial resolution MSR 3 Imaging spatial resolution ISR Point visibility factor PVF is defined as normalized centre pixel signal caused by a point source It is calculated as ratio of the centre pixel signal to the sum of signals generated by the point source in both the centre pixel and the neighbor pixels PVF center pixel signal sum of all pixels Ge PVF can be determined using the four step algorithm presented below 1 Capture an image of a uniform background Frame 1 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 Fig 3 28 Exemplary signal distribution of uniform background image 2 Capture an image of a point source on the uniform background Frame 2 100 100 100 100 100 100 108 130 109 100 100 130 210 128 100 100 120 131 109 100 100 100 100 100 100 Fig 3 29 Exemplary signal distribution of point source image 3 Calculate a new frame as the difference of Frame 1 and the Frame 2 0 0 0 0 0 0 8 30 9 0 0 30 110 28 0 0 20 31 9 0 0 0 0 0 0 Fig 3 30 Signal distribution after backgr
108. e to recalibrate the 162 test system in local conditions In most cases recalibration of such blocks like blackbody or collimator can be done in local metrological centers assuming that some technical information is revealed by the manufacturer Sixth typical twelve month warranty time is quite short Extended warranty of ten means well spent funds as such warranty guarantee full manufacturer respons ibility in much longer time Seventh you will spend well your time by reading the educational section at websites of manufacturers of test equipment or by reading other available specialized literature It will be later easier for you to communicate with the manufacturers because you will then know exactly what you need Eight it is possible to built a system for testing thermal imagers by buying some blocks collimator blackbody targets and developing other needed blocks rotary wheel control software test software However it is a risky policy recommended only for technologically advanced _ scientific manufacturing centers with deep knowledge of metrology of thermal imaging Probability of commercial losses due to long time needed for development of such a combined test system and due to possible technical problems is high in most cases Ninth systems for testing thermal imagers offered on international market by different manufacturers are generally similar The differences are created by small technical details Some of these detail
109. ed although multiple observers are recommended 1 If positive and negative contrasts have been measured the individual MRTD 2 7 2 The MRTD results shall be tabulated see Annex and plotted on a graph with a logarithmic scale for temperature dif ferences and a linear scale for spatial fre quencies for each observer and each ori entation The measurement conditions values used for calculation and all the res ults must be shown Any significant de viation or rejection of results must be ex plained 161 equipment Manufacturers of test equipment offer typically a series of at least twenty standard 4 bar targets of different spatial frequency Next they can optionally deliver cus tom designed 4 bar targets of spa tial frequencies outside the list of standard targets Such a situation gives users of test system possib ility to fulfill requirements of point 6 4 on targets used for MRTD measurement Test control software uses the for mula recommended in point 7 1 The measurement results can be recorded using the test software in several typical data formats MS Excel Later the data can be presented graphically using any software for data presentation 7 Guidelines for buyers of test equipment Systems for testing thermal imagers are built from at least six blocks collimat or blackbody rotary wheel set of targets PC frame grabber software Blocks of the system for testing thermal imagers were
110. ed within the spectral range 1 3 um It is questionable whether the SWIR cameras are thermal cameras as in this spectral range the reflected radiation dominates over the emitted radiation for the objects of temperatures below about 100 C However let us treat them as a group of thermal cameras because of very similar design to MWIR and LWIR thermal cameras At present the SWIR cameras are only a marginal group of thermal cameras The SWIR cameras have been commercially available on the market for no more than a decade This situation originated the fact that the SWIR range has not been an interesting range for both military and civilian applications for many decades Due to dominance of the emitted thermal radiation and the atmospheric windows military agencies were interested mostly in the MWIR and LWIR ranges Because of sensitivity range of human sight and well developed silicon technology the civil ians were interested in the visible and NIR ranges This lack of significant interest created the situation when up to the middle of the 1990s no well matured technology of detector arrays for SWIR range was available 1 Currently this vacuum is occupied by InGaAs arrays and the SWIR cameras have found a number of applications The SWIR imagers are quickly gain ing popularity in surveillance market due to higher image resolution than MWIR LWIR thermal imagers and better performance at limited visibility condi tions than typical TV cameras based on
111. els but after correction of spatial noise NETD is now a measure of average temporal noise for a group of pixels Both three methods can be used with or without correction of low frequency trends This means that depending if we use the correction or not then NETD can be a measure of only high frequency noise or of a measure of full bandwidth noise To summarize NETD can be treated as a useful metric of a noise phenomenon but only if it is precisely known how it was measured without this knowledge NETD data can be very misleading The differences in NETD values get using dif ferent calculation methods can be very significant NETD measured using the first method can be several times lower than NETD measured using the second method This situation is particularly drastic in a field non cooled thermal cameras when the total noise is typically a several times higher than its temporal component 3 3 3 Four parameters approach Noise phenomenon is too sophisticated to be precisely characterized by a single parameter like NETD Even if it is clearly defined how NETD was determined we get too little information about the noise phenomenon Two thermal cameras can have the same NETD defined using any of different methods presented earlier but a human eye will immediately notice big differences in images generated by these thermal cameras On the other hand as it was stated earlier 3D noise model using eight components for precise characterization of
112. ement is done at longer distances or at poor atmospheric conditions then atmospheric transmittance should be calculated and its influence on measurement results corrected The correction is done using the formula presented below MRTDen MRTD Ta 5 4 where 1 1s the atmospheric transmittance Table 5 1 The correction coefficient cor for typical thermal imagers of the spectral bands 3 5 um and the 8 12 um at different ambient background temperature Temperature Non cooled Non cooled Cooled Cooled Toa 3 5 pm 8 12 wm 3 5 pm 8 12 pm 0 0 55 0 78 0 74 0 88 2 0 58 0 8 0 76 0 89 4 0 62 0 82 0 79 0 91 6 0 66 0 84 0 81 0 92 8 0 7 0 87 0 84 0 93 10 0 74 0 89 0 86 0 94 12 0 79 0 91 0 89 0 95 14 0 84 0 93 0 91 0 97 16 0 89 0 96 0 94 0 98 18 0 94 0 98 0 97 0 99 20 1 0 1 0 1 0 1 0 22 1 06 1 03 1 03 1 01 24 1 12 1 05 1 06 1 03 26 1 18 1 07 1 09 1 04 28 1 25 1 1 1 12 1 05 30 1 32 1 12 1 15 1 06 32 1 39 1 15 1 18 1 07 34 1 46 1 17 1 21 1 08 36 1 54 1 2 1 24 1 09 38 1 62 1 22 1 27 1 1 40 1 7 1 24 1 3 1 12 5 1 4 Targets MRTD is a continuous function of temperature difference on target spatial fre quency However MRTD is measured only at several points The results are later approximated and MRTD curve is presented It is typically considered that MRTD should be measured at at least three measure ment points that area located at three spatial frequency ra
113. enient than using control ler keyboard as the user has at his disposal a large keyboard a mouse and a monitor of the PC instead of a small keyboard and a small screen of the blackbody control ler Next the control software enables the user also visualization and recording temperatures versus time and recalibration of the blackbody The presented above description block diagrams and photos refer precisely to old model of blackbodies offered by Inframet 30 New models from this company look a different but basically the work concept is the same as presented earlier 4 2 2 2 Requirements Temperature resolution NETD of modern cooled thermal cameras can be as low as about 10 mK Temperature resolution NETD of uncooled thermal cameras is worse typically about 100 mK but is improving rapidly If we want to test accur ately thermal imagers then we need blackbodies of temperature resolution a dozen or more times better than imager temperature resolution NETD Temperature res olution of 10 mK is acceptable in case of non cooled thermal cameras However if cooled thermal imagers are to be tested then blackbodies of 1 mK resolution are needed In some extreme cases when imager MRTD at low frequency range is be low 10 mK then even 0 1 mK resolution is useful Differential temperature range 10 C at 25 C ambient temperature can be con sidered as acceptable during tests of typical surveillance thermal imagers However wider temperature range is u
114. er 3 Thermal imager is looking into a medium size active blackbody a bit big ger than imager optics located at very short distance from imager optics 4 Thermal imager is looking into a passive large blackbody located at short distance to the imager optics 5 Thermal imager optics is covered using a cap black cloth or other materi als Method number one is the most easiest to be used The needed blackbody is a part of typical test systems However field of view of tested thermal imagers is usually much bigger than angular size of image of the blackbody projected by the collimator Therefore analysis area is limited only to a part of images generated by thermal imagers and measurement using this method generates significantly lower value of spatial noise measurement Temporal noise is sometimes also lower than true measurement results for full field of view The disadvantage of the methods no 2 3 is the fact that an additional large size medium size blackbody is needed for noise measurement Another option is that the blackbody used as a part of the variable target test system if the blackbody size is big enough to cover the imager optics must be moved to a new position very close to the optics of the tested imager 151 For method no 4 a large passive blackbody of non regulated temperature that keeps ambient temperature is needed An advantage of this method is lower manu facturing costs than in case of methods 2 3 The method no
115. er and the term thermal camera will be used as equivalent terms 2 2 Applications of thermal cameras According to their applications thermal imagers can be generally divided into two basic groups surveillance thermal cameras and measurement thermal camer as The surveillance thermal cameras are mostly used in military applications for observation of a battlefield in darkness or in difficult atmospheric conditions by cre ating the relative temperature distribution of the terrestrial scenery being observed The measurement thermal cameras are used for civilian applications in industry and science mostly for non contact measurement of temperature distributions on the surface of the tested objects Nowadays the borderline between these two groups becomes more fluid as there are some cameras that can be used for both ob servation and measurement applications However this situation is still an excep tion from the rule as most surveillance cameras do not have capabilities to measure temperature of the observed objects and the image quality of the measurement sys tems is inferior to the image quality of the observation thermal cameras Image quality is the most important criterion for performance evaluation of sur veillance military thermal cameras In case of measurement commercial thermal cameras the situation is more complicated Applications of measurement thermal cameras can be divided into two general groups the applications
116. er b the thermal imager generates an image of the target of interest but we cannot determine accurate relative absolute coordinates of the target Boresighting of thermal imagers can be done using special versions of systems for testing thermal imagers or using specialized systems optimized for boresighting task There are generally three types of systems for boresighting of thermal imagers or multi sensor surveillance systems The first systems are based on a concept of a collimator generating output beam in direction of exactly known angular coordinates The second systems are based on a concept of an additional imaging sensor fixed to a reference mechanical axis looking into direction of the test collimator The third systems use an additional imaging sensor fixed to a reference mechanical axis looking into infinity target real target at long distance Boresighting process is generally outside area of interest of this book and shall not be discussed further More detailed information about equipment for boresighting of thermal imagers can be obtained from manufacturers of equipment for testing thermal imagers who can optionally deliver also boresighting equipment 4 4 References 1 ASTM standard E 1213 2002 Standard Test Method for Minimum Resolvable Temperature Difference for Thermal Imaging Systems 2 ASTM standard E 1311 99 Standard Test Method for Minimum Detectable Temperature Difference for Thermal Imaging Systems 3
117. er the MSR is usually a few times higher than ISR and manufacturers often prefer to present only the values of the second para meter 3 4 4 Subjective parameters based on resolution targets There are myriads of resolution patterns developed during last century to measure resolution of optical instruments and later to measure resolution of image intensifier systems or television cameras The USAF 1951 target the EIA Tn the presented above exemplary case we get PVF 0 4 62 Resolution Chart 1956 the NBA 1963A Resolution Target can be considered as the most popular targets from this group These resolution patterns can be potentially also used to characterize resolution of thermal cameras However there are some technical problems that make difficult direct use of commercially available resolution targets Typical resolution targets offered on the market 32 are manufactured using two techniques a translucent targets chrome pattern on glass substrate b opaque targets printed black pattern on mylar high quality paper The problems is that typical glass substrate does not transmit in spectral bands of thermal imagers Next emissivity of printed patterns can be similar to emissivity of the mylar paper These reasons create a situation when thermal cameras do not see typical resolution targets and such targets cannot be used for testing thermal cameras A new type of USAF 1951 targets was recently introduced on commercial mark
118. er no 2 controls the keyboard and enables the user to set the required absolute or relative temperature Next the micro controller 2 controls the screen and enable visualization of current and required values of temperature of the emitter plate or the temperature difference between the emitter temperature and the front wall temperature 95 Task of the power supply module is to power the micro controller 1 module the micro controller 2 module and the Peltier thermoelement and analog electronic module of the radiator block a b Fig 4 16 Photo of the TCB 2D blackbody a RTCB radiator block b CTCB controller block courtesy of Inframet The main task of the controller is to control and stabilize absolute or differential temperature of the emitter in the radiator block Stabilization is a temperature con trol process of the emitter with aim of achieving its stable temperature equal to the value set by the user Temperature control of the emitter plate is done via preci sion control of the voltage applied to the Peltier element The easiest way to shorten time necessary to achieve the stable temperature ofthe emitter plate is to decrease thermal capacity of the emitter plate However lower thermal capacity means also lower thermal uniformity of the emitter and therefore the only reasonable way to shorten stabilization time is to optimize control algorithm of the voltage applied to Peltier element Classical PID propor tional integra
119. er to carry out semi automatic MRTD measurement it is necessary to carry out first classical MRTD measurement of a few imagers of the same type in order to determine correction coefficients The coefficients are valid only for just this type of thermal imagers This requirement makes the semi automatic MRTD measure ment attractive only when large number of imagers of the same time is to be tested 116 5 1 1 Test team MRTD is a subjective measurement when decision about test results is made by a human observer It is natural that there is some variability of human sight within human population In order to correct this variability both standards and specialized literature recommend to carry out MRTD measurement by several at least three observers and then to average test results 1 4 In this way the variability of results get by different test teams should be theoretically eliminated Practically differences between MRTD measurement results of the same thermal imager obtained by several test teams at the level as high as 50 are quite common 2 3 There are two basic reasons for this situation First differences in test equipment detection criterion observation conditions and in a measurement procedure used during testing Second training level of the test crew There is a general tendency that observers who spend hundreds of hours working with thermal imagers are well adapted to typical noisy thermal images and they get better results
120. eral ambient temperatures using a temperature chamber but this recommendation can be treated as an optional one Calibration of measurement thermal imagers could be potentially carried out using a typical system for testing surveillance thermal imagers when the tested measurement imagers is looking into the blackbody via the collimator However practically blackbodies from systems for testing surveillance imagers can be used only for partial calibration of measurement thermal imagers The absolute temperature range of typical blackbodies used for testing surveillance range from 112 0 C to 100 C is too narrow for calibration of measurement thermal imagers designed for non contact temperature measurement of targets of temperature variable from about 20 C to about 300 C sometimes up to 1000 C There are on the market some differential blackbodies of extended temperature range from about 15 C up to about 150 C 180 C 24 30 Such blackbodies can be used for calibration even more for recalibration of measurement thermal imagers optimized for temperature ranges no more than only slightly bigger than the absolute temperature range of these blackbodies However upper limit of temperature range of most thermal imagers is typically much higher than 180 C Therefore several blackbodies of wide combined temperature range are needed to carry calibration of most measurement thermal imagers temperature chamber black thermal body imager
121. ermographica 1978 32 Zhu D et al Temperature measurement of gas flow with high temperature and speed using thermal video system Proc SPIE Vol 3783 p 373 379 Optical Diagnostics for Fluids Heat Combustion and Photomechanics for Solids Soyoung S Cha Peter J Bryanston Cross Carolyn R Mercer Eds 10 1999 17 2 Review of thermal imagers Thermal imagers are imaging systems that generate images of the observed scenery using thermal radiation emitted by the scenery These numerous imaging systems can be divided into several different groups First according to a method of creation of two dimensional image of the ob served scenery thermal imagers can be divided into two distinct groups thermal cameras and imaging thermal scanners Second according to application area thermal imagers can be divided into two groups surveillance thermal imagers and measurement thermal imagers Third according to a spectral band thermal imagers can be divided into two optionally three groups MW mid wave infrared thermal imagers and LW long wave imagers Sometimes SW short wave thermal imagers are added too Fourth according to technology of IR detector IR FPA thermal imagers can be divided into at least three different generations Fig 2 1 Classification of thermal imagers 2 1 Thermal cameras versus thermal scanners Thermal camera is a thermal imaging system that enables us creation of a two dimensional thermal ima
122. erture collimators can be used aperture of about 100mm F number ratio of focal length to aperture of collimating mirrors used in IR collimators vary from about 5 low F number collimators to about 12 high F number collimators This means that focal length of collimators of the same aperture but of different F number will vary significantly For example in case of 150 mm aperture collimators of variable F number from 5 to 12 the focal length varies from 750 mm to 1800mm There are advantages and disadvantages of both types of the collimators Low F number collimators are characterized by small size that enables to decrease dimensions of the complete test system However thermal stability of low F number collimators is lower than in case of high F number collimators Next manufacturing mirror accuracy is usually better in case of high F number mirrors However the most important disadvantage of low F number collimators are significant geometrical aberrations that occur for off axis spots Fig 4 13 This means that they project a perfect image of a small spot 4 bar target located in the center of a collimator field of view but an image of several 4 bar targets located outside the center can be blurred 88 So foe NI s oD O s aberration blur mrad S P hS s 0 5 10 15 20 2 30 35 4 45 offfocus distance mm Fig 4 13 Aberration blur versus off axis distance for several reflec
123. es and rods on the retina is not uniform The highest concentration of the cones is in the retina s center much less concentration on the peripheral parts of the retina In case of the rods we have an inverse situation Therefore in daylight condition a human eye has the highest acuity in the center of its field of view in the night conditions at periphery of the field of view The principle of seeing phenomenon is the following The light emitted by the Sun or other light sources illuminates the observed scenery and it is reflected The reflected light carries information about the targets that reflected the light Different targets of the scenery differ in their ability to reflect the incoming light The targets of high reflectance will be perceived as bright ones the targets of low reflectance will be perceived as dark ones Because reflectance depends also on a wavelength then the reflected radiation carries also information about color of the observed target To simplify the target of high reflectance in the spectral band of 0 65 0 78 um and low reflectance in the spectral band of 0 4 0 5 um will be perceived as a red one In case of inverse situation the target will be perceived as a blue one The image generated by the eye lens must be created exactly on the retina The distance between a human eye an observed target can vary significantly In order to compensate this distance variation the focal length of the lens must vary This
124. esponse than any artificial light sensing device and high power to adapt to different illumination conditions It is characterized by high accuracy of alignment and it can distinguish between hundred of thousands of colors or tones It can operate the best in daylight condition but also works relatively well in weak illumination conditions However a human eye is not perfect and there are a few serious disadvantages of human eye from the point of its effectiveness as a surveillance tool 1 Limited spatial resolution that limits perceiving small details of the targets particularly long distance targets 2 Limited sensitivity that decreases effectiveness of observation in night condi tions 3 Short surveillance ranges in bad visibility conditions in a visible range fog rain snow dust 4 No capabilities to record images to process them and to transfer to other hu mans in easy accurate way 5 Relatively easy camouflage in a visible range 6 Human eye is sensitive only in a narrow spectral band of optical radiation i e in the visible range Electro optical imaging systems improve capabilities of a human eye and remove at least partially the mentioned above limitations 1 3 Division of electro optical imaging systems From a point of view of a support to human eye there are basically three types of electro optical imaging systems night vision devices night vision devices TV cameras and thermal imagers Night vision devices NVD
125. est systems The variable target distance test systems represent a mixture of two earlier presented test concepts Their main advantage is lower costs and lower mass due to elimination of the col limator in comparison to variable target test systems The disadvantage is the fact that the systems can be used only for testing thermal imager of very short minimal focus range like typical commercial measurement thermal imagers Summary comparison of the presented three groups of test systems for testing thermal imagers is presented in Table 4 1 Table 4 1 Comparison of different types of test systems Variable target test Variable distance test Variable target distance systems systems test systems Design IR_ collimator rotary large blackbody set of rotary wheel small structu wheel small blackbody large targets shield blackbody set of small re set of small IR targets box portable PC IR targets shield case PC frame grabber frame grabber PC frame grabber software software software Advanta classical mature design compact design lower cost than in case ges high measurement possibility to use of variable test system accuracy all types ofi outside laboratory low compact design thermal imagers can be cost tested Dis difficulties to use Large rooms corridors suitable only for imagers advanta outside laboratory needed for indoor of short focus distance ges limitation on
126. et so called the Clear Optical Path USAF target 32 Such targets are manufactured from an extremely thin electroformed nickel substrate Since there is no glass in the pattern area light travels only through air eliminating chromatic and absorption issues and can be potentially used for testing thermal imagers However the targets are characterized by too low emissivity about 0 7 0 8 in comparison to typical IR targets of emissivity over 9 96 Next the reflectivity of such targets is not very high about 0 7 0 8 and the targets absorb radiation emitted by the blackbody located behind the target in systems for testing thermal imagers This radiation can change temperature distribution on the surface of the new type USAF 1951 target Because of these reasons the Clear Optical Path USAF targets should not be used for MRTD measurement of thermal cameras However these targets can be used for relative comparison of resolution of different surveillance thermal cameras 3 5 Accuracy parameters So far we have discussed parameters useful for testing and evaluation of surveil lance thermal cameras In case of this type of thermal cameras image quality is the most important figure of merit However in case of measurement commercial thermal cameras high quality thermal image is useful but accuracy of non contact measurement of temperature is more important There are two types of errors of a temperature measurement with thermal camer as t
127. evaluate properly thermal cameras not only image quality detection recognition and identification ranges but also other factors like mass dimensions resistance to harsh environ mental conditions ergonomics must be taken into account Further on there are on the market thermal cameras integrated with additional modules like GPS laser range finder goniometer day light TV camera and laser pointer These additional modules can significantly increase capabilities of a thermal camera To summarize evaluation and comparison of thermal cameras is a complicated and risky task that requires to take into account a set of factors that could vary depending on the final user needs 24 Fig 2 6 Sophie MF thermal camera integrated with laser range finder goniometer day light TV camera and laser pointer courtesy of Thales Optronique Detectors used in Gen 1 Gen 2 and partially Gen 3 of thermal cameras require cooling typically to the temperature equal to 77 K First thermal cameras were cooled using dewar coolers The dewar cooler is essentially a vacuum bottle filled with a coolant Different liquid gases can be used as coolants However liquid nitrogen is used as a coolant in almost all dewars used in practice The cryogenic cooling is characterized by a few significant disadvantages like ne cessity to have a source of liquid nitrogen supply readily available limited working time of the dewar after filling and necessity to
128. f the responsivity function It is calculated as tangent of the angle between linear part of the responsivity function and the temperature axis the slope of the linear part Blackbody temperature in mK is typically treated as the input signal and the screen luminance in digital gray scale levels is treated as the output signal The formula used for SiTF calculation is shown below poe oT SiTF TEER 5 18 where S and S are the output brightness of the monitor in grey level units generated by the input target temperatures T and 7 gt The values of T and T must be in the linear range of the responsivity function Practically knowledge about responsivity function or responsivity functions in most cases is useful for several reasons First analysis of S shape of the responsivity function deliver precise informa tion if the electrical analog video signal is properly set case of imagers generating output image using PAL NTSC formats A situation when output signals in the range from about 0 to 255 are generated by target of temperature in the required temperature range is usually desired The required temperature range depends on geographical regions If the video signal is not properly set then imager dynamics can be lower than expected In other words we can have a situation when image of even very cold targets shall not be truly black and image of a very hot target is not white It is also possible to have a situation when the i
129. f the edge target after noise reduction process MTF of the tested imager is calculated by the test software 8 Storing the MTF measurement results in form of table text file or a graph ics 5 2 3 Interpretation of measurement results Accurate MTF measurement of thermal imagers is not easy The same can be said about interpretation of the measurement results as there are several factors that can influence measurement accuracy First measured MTF always depends significantly on proper focusing of tested thermal imager Therefore the focusing stage in the MTF measurement procedure should be done very carefully 146 Second MTF measurement results depend often on setting of the tested imager like gain level or image processing Therefore write down always measurement conditions Gain Brightness Image processing features etc used when MTF measurements were carried out As we can see in the graph below MTF of the same imager can be drastically different when measurements are carried out at different imager settings Therefore please compare several thermal imagers only if the im agers were tested having the same settings It is typically recommended to have the following settings during MTF measure ment Gain minimal Level medium However for low Gain in order to achieve reasonable contrast between the target and the background temperature difference at level over 2 K are needed There are however on the market cooled IR FPAs
130. ge of the observed scenery independently whether the sys tem or objects are movable or stationary ones Imaging thermal scanner is a thermal imaging system that provides creation of a two dimensional thermal image of the observed scenery only when the scanner or the objects are moving Thermal cameras represent probably over 99 of all existing thermal imagers Imaging thermal scanners are almost exclusively airborne systems used for recon 18 naissance applications because they offer very wide field of view standard 120 in contrast to the thermal cameras offering field of view not wider than about 30 Because of distinct differences in design of these two types of thermal imaging sys tems and narrow specialized market the imaging thermal scanners are very expens ive systems Due to mass application of thermal cameras their prices are signific antly lower There exist numerous literature on both imaging thermal scanners and on thermal cameras We can here only mention that detail presentation of a design of thermal cameras was presented in Refs 6 7 12 11 10 and detail dis cussion on a design of imaging thermal scanners in Ref 5 As we mentioned earlier thermal cameras are the most numerous group of thermal imagers Practically almost all thermal imagers are thermal cameras Therefore both review of thermal imagers and later analysis of test methods in next chapters is mostly limited to thermal cameras Next the terms thermal imag
131. generated by optical and electronic systems Therefore the target transfer function TTF calculated as the ratio of the real ATF to the ideal ATF provides useful in formation about this phenomenon The point visibility factor PVF is a point of TTF function determined for the conditions when the target area approaches zero The PVF is also sometimes termed the en squared energy EE or en squared power EP or blur efficiency ATF IFOV Area Fig 3 14 Ideal and real aperiodic transfer function ATF IFOV Area Fig 3 15 Target transfer function 43 3 2 3 Slit Response Function The SRF is defined as a function of the signal generated by a slit versus width of the slit normalized to the signal generated by a very wide slit The SRF can be treated as one dimensional ATF SRF generally provides directly information on the system ability to detect long narrow targets The SRF function is used as a base to determine so called measure ment resolution and imaging resolution The latter parameters shall be discussed later SRF 1 00 IFOV angle mrad Fig 3 16 Slit Response Function 3 3 Noise parameters Noise is a phenomenon that can significantly decrease image quality and limit system ability to detect low contrast targets Noise parameters are very important performance measures of thermal cameras Noise present in thermal images can be in general divided into two groups tem poral noise and spatial noise
132. gers can be found in international standards that regulate testing thermal imagers 1 2 17 19 However real value of this information is rather limited due to small size of these standards lack of many important details some out dated re commendations More useful information can be found in some specialized literat ure sources like Refs 10 11 However the most valuable information about test equipment can be get from analysis of real test systems offered by manufacturers of equipment for testing thermal imagers 24 26 29 30 32 Here in this chapter con clusions from such an analysis are presented Due to limitations in information ac cess most detail technical information refers directly to test equipment from a single manufacturer Inframet 30 However there are no basic differences in test systems from different manufacturers only minor technical details All properly working test systems independently of the manufacturer should generate the same meas urement results Therefore the information about equipment for testing thermal im agers presented in this chapter should be generally valid for all test systems avail able on international market 4 1 Types of test systems The task of a test system for testing thermal imagers is to generate images of some standard static targets of precisely known shape dimensions and temperat ure These images can be projected to the tested thermal imager by the test system or viewed directly by
133. ges of the gain and the offset of these blocks Influence of the environment temperature on the measurement results can be corrected Modern thermal cameras are equipped with software and hardware that should automatically correct this influence However only a partial correction of this harmful influence is possible Therefore accuracy of measurements carried out in real measurement conditions can differ significantly from accuracy obtained in laboratory conditions There are also several parameters presented in the catalogs of measurement thermal cameras that give some indications about intrinsic errors of thermal camer as thermal sensitivity IFOV and rarely MRTD or MDTD 64 99 Ge Thermal sensitivity called also thermal resolution temperature resolution or NETD that provides information about the influence of noise in electrical channel on the measurement errors It was shown in Ref 6 that NETD equals the standard deviation of the output temperature dispersion caused by noise of the sys tem Therefore the NETD can be treated as a good estimation of uncertainties due to system noise However we must remember that the NETD depends on an object temperature It is typically measured only for one fixed value of this temperature usually close to 30 C and can be a few times higher for object temperatures at the lower limit of available temperature range close to 20 C The instantaneous field of view IFOV is typical
134. ges of effective surveillance ee eeeeeeeeeeeeceneeteneeeneenes 36 3 2 RESPONSE PAFAMClETS ssccscssessccessecencsuassduncenceseasessesstansendescendensdeskasatesesteswesdedsesseues 40 3 2 Responsivity finiction 3 xen n Re ah Bes aes 40 3 2 2 Aperiodic Transfer FUNCtION ccccceceescesseeseesseeseeeseeseceseseeeaeeeenseesenaeeeeenaes 42 3 2 3 Slit Response FUNcCtion ccceccescesceseeseeeseeseceseeseceseescessecseeeseceeeeaeessesseensaees 44 3 3 Noise parameters i vicciccscdssesssecscssscevcsseonddeseseovssnassivsceieaseassevesssendescesebossessSensensessbeas 44 3 35 231 noise mod l ienee e e a eae a ES 45 3 3 2 Single parameter approach ssssessesssseesessesessessrsestsresessesesessestesessrssreessessse 47 3 3 3 Four parameters approaches a E ERE 48 3 4 Image resolution parameters eeseesseesoesoessossossocseossosscossossosssoesossoossossossoosseseoossss 54 3 4 1 Parameters based on specifications of the IR FPA 0 cecceeesceeseeteeeeteeeenees 54 3 4 2 MTF and derivative parameters cccscccsesseesecseeeseeeseeeeseeeeeeeeeesneeeesteeeneaees 56 iv 3 4 3 Parameters based on imager response to point slit sources ceeeeereeee 61 3 4 4 Subjective parameters based on resolution targets c ecceesesseeteeeteeteeeees 62 3 5 ACCULACY PALAMEte S csscsrccsscerscsscersssscessscssessccsesssessssseesssssssssessessssssseceses 63 SiO SUMIMANY 35s cccccesssciecsncscsccasesceoussev
135. get is used for measurement of distortion The distortion effect in thermal imagers of narrow medium field of view is usually small almost negli gible However the same target can be used also for measurement of imager field of view It is important to measure the latter parameter because only when we known field of view there is any sense to compare MRTD characteristics of several thermal imagers It may be surprising for some readers that square targets are not included in this basic set of IR targets The square targets were traditionally used for several dec ades during measurement of NETD of thermal imagers However it should be noted that these targets were used to measure NETD of scanning thermal imagers using oscilloscopes Nowadays frame grabbers are typically used instead of os cilloscopes Next staring thermal imagers are typically tested instead of scanning imagers Finally modern software enables analysis of any part of images generated by thermal imagers Therefore the rectangle targets are not strictly needed for meas urement of noise parameters These targets are needed for measurement of ATF but it is a rather rarely measured characteristic Other types of IR targets listed in Table 4 7 are needed in case of extended tests of thermal imagers or when by use of multi pattern targets we want to speed up test procedure Applications of these targets are presented in the same table 4 2 5 Image acquisition analysis module Image
136. h value of probability From analysis of the presented above definitions we can draw three basic con clusions Because the term measurand is relatively new and still not accepted widely in literature the term measured quantity will be used in the rest of this book 12 First that accuracy is only a qualitative concept that should not be associated with numbers This means that we should not specify instrument accuracy as equal to a certain number as it is unfortunately a common practice so far We are allowed according to the VIM to say only that accuracy is good bad etc Second the defined according to the VIM term error is a perfect measure of measurement accuracy However this true error of measurement is always un known because the true value of the measured quantity is unknown The same can be said about its component the systematic error Let us temporarily call the term error as the true error to make a better distinction with the term limit error Third two other measures of measurement accuracy the uncertainty and limit error of the result of a measurement may be evaluated These two measures of measurement accuracy are useful for users of measuring instruments who know only the instrument indication and want to estimate accuracy of the measurement result Guidelines on evaluation of uncertainty of measurement results are presented in the mentioned earlier Guide to the expression of uncertain
137. hanical axis 4 3 1 Correction of spatial noise Image generated by modern IR FPA sensors particularly non cooled sensors is typically very noisy Spatial noise is the dominant type of the noise Correction of spatial noise is needed to obtain a clear image generated by thermal imagers There are many algorithms that are used to correct spatial noise in thermal im agers However they are generally based on a concept of IR FPA module thermal imager looking into an internal external uniform radiation source Fig 4 24 We can say that in order to correct spatial noise we must create a data array that records values of this noise at different ambient temperatures and at different input signal levels Practically such data array is determined by putting a tested IR FPA module or tested thermal imager located at a very short distance to an area blackbody to a temperature chamber and by recording images at different chamber temperature for different blackbody temperature and for different imager settings gain level etc This operation provides primary data base used for spatial noise correction Recalibration of thermal imagers using internal mechanical chopper provides only additional secondary data base Some manufacturing teams carry out corrections of spatial noise on the basis of data obtained during tests of IR FPA modules others carry out corrections of spatial noise on the basis of data get during tests of complete thermal imagers
138. have collimator influence on degradation of image generated by tested thermal imager negligible then the collimator resolution Veo must be at least 5 times better than the thermal imager resolution Vy Vegi O Yy 4 2 The spatial resolution vy defined as Nyquist frequency of the thermal imager can be calculated using data provided by the manufacturers using this formula v aa ee e 43 N 2 FOV mrad oe where N is the number of pixels in horizontal or vertical direction of FPA used in imager design FOV is imager field of view in horizontal or vertical direction The spatial resolution of the collimator V cannot be calculated but it can be measured It can be done using a measurement method that was proposed in Ref 5 The method is based on an idea to carry out measurement of collimator spatial res olution in visible spectral range because geometric aberrations of typical reflective IR collimators do not change with a spectral range It is also extremely important that the measurements should be carried out not only when the resolution targets located exactly in the collimator focus point but in several position located within a circle around the focus point It is suggested that the diameter of such the circle should be at least 14 mm the area wide enough for a typical multi pattern target If such tests are carried out then we get information about real capabilities of tested collimator to project high quality images 85
139. he external errors and the intrinsic errors 4 Here we will present parameters of measurement thermal cameras that describe camera performance when only the intrinsic errors are present Such a situation occurs when emissivity of the tested object is close to unity and a distance camera object is short Then the external errors due to unknown emissivity reflected radiation and limited atmo spheric transmittance can be treated as negligible Manufactures of measurement thermal cameras often state a parameter called accuracy that is measured as a range around the true object temperature Ta in 63 which the output temperature Tou is located when the external sources of errors are negligible Typical values of this parameters are 1 of the output temperature Tou but not less than 1 C for scanning thermal cameras or 2 of the output temper ature Tou but not less than 2 C for staring thermal cameras The term accuracy according to the international metrological organizations is only a qualitative concept that should not be associated with numbers 11 There fore the name accuracy is improper for formal metrological terminology How ever there are more serious limitations of usefulness of the accuracy parameter The accuracy parameter could potentially enable determination of the intrinsic uncertainty of a measurement thermal camera Assuming a uniform distribution dis persion of a true temperature within the
140. hermal camera y can be presented in different signal units digital levels volts etc However as long as SiTF is expressed as a ratio of the same unit divided by K then NETD is calculated in Kelvin degrees As it was presented above the situation with NETD parameter as a metric of a noise phenomenon is relatively clear in case of scanning thermal cameras It is a measure of high frequency temporal noise in a single line Situation is much more complicated in case of array thermal cameras Different definitions of NETD are often used NETD is still defined as a blackbody temperature difference between a target and its background required to produce a peak signal to ems noise ratio of unity A sequence of images of uniform target is typically used as raw data for NETD calculations However the difference depends on the way how to calculate this ems noise NETD can be calculated in case of staring thermal cameras using different methods First ems noise can be calculated as a standard deviation of temporal variation of a signal of a single pixel NETD is now a measure of only temporal noise of a single pixel Second ems noise can be calculated as a standard deviation from temporal and spatial variations of signals from a group of pixels NETD is now a measure of total noise both temporal and spatial noise components Third ems noise can be calculated as a standard deviation from temporal and spatial variations of signals from a group of pix
141. hermal imagers are electro optical imaging systems sensitive to mid wave and long wave infrared radiation that generate images of the observed scenery us ing thermal radiation emitted by the scenery They have found numerous applica tions in both defence amp security sector military border guards police etc and ci vilian sector industrial non contact temperature measurement non destructive thermal testing tests of electrical power lines building industry medical applica tions fire rescue etc A lot of myths about thermal imagers are known There is a rich literature on subject of thermal imaging but there are also papers presenting conflicting conclu sion about performance of these modern fascinating imaging systems Different rules how to choose an optimal thermal imager can be found in available literature too It is also known that quality of thermal imagers offered on the market vary sig nificantly The only way to be sure about quality of imagers of interest is to test them and to evaluate test results Testing thermal imagers is a very difficult task Extended knowledge from dif ferent areas like physics optics electronics thermal sciences precision mechanics metrology and practical experience with thermal imagers is needed to carry out ef fectively testing modern thermal imagers There is rich literature on the subject of testing and evaluation of thermal imagers and there are also several standards that regulate tests
142. iable test system two 4 bar patterns of different bar width 8 mm and 4 mm were used Bigger 4 bar pattern 8 mm bar width was used during measurement at low frequency points smaller 4 mm bar width at high frequency range This solution was used in order to minimize the distance needed to carry out MRTD measurement at the required spatial frequency range from 0 5 Ip mrad to 4 Ip mrad Table 5 11 Exemplary table with vertical MRTD test results obtained using a variable distance test system Temperature difference K Spatial frequency lp mrad MRTD MRTD or Vv vV V3 Vay 0 0120 0 0118 xX X x x 0 025 0 0245 X x X x 0 05 0 049 0 51 0 5 0 49 0 49 0 1 0 098 1 03 1 0 96 0 96 0 2 0 196 1 55 1 51 1 45 1 45 0 4 0 392 2 06 2 02 1 97 1 97 0 8 0 784 3 2 92 2 95 2 95 1 6 1 568 3 54 3 52 3 4 3 4 139 0 01 spatial frequency lp mrad Fig 5 9 MRTD measurement results obtained using two methods squares variable target meth od triangles variable distance method 5 1 10 Software support The MRTD is a subjective parameter that describes ability of the imager human system for detection of low contrast details of the observed object The human is to make decision when he can resolve the thermal image of the bars Therefore com puter technology is not generally needed during MRTD measurements However software support can still be useful during MRTD measuremen
143. ical that the tests of thermal imagers are typically limited to measurement of only MTF function Fifth parameters used to characterize surveillance thermal imagers differ from parameters used to characterize measurement thermal imagers Therefore typical test teams that specialize in testing surveillance thermal imagers do not need to know definitions and test methods of parameters of measurement imagers like accuracy NGE thermal stability MSR Sixth performance of thermal imagers do not depend significantly on spectral parameters At the same time these parameters do not typically vary with time Therefore the spectral parameters are only rarely measured by test teams of manufacturers of thermal imagers very rarely by users of these imagers To summarize detail knowledge about all parameters of thermal imagers discussed in this chapter is not needed for both the users and the manufacturers of thermal imagers Optimal set of parameters to be measured depends on potential use of measured parameters In case of users of surveillance thermal imagers the tests are generally carried out to verify potential performance of tested imagers Due to direct relationship of MRTD parameter and ranges of effective surveillance the tests are often limited to measurement of this parameter For manufacturers of thermal imagers the aim of the tests is not only to verify performance of the imagers but also to find weak spots of the tested imager that could be po
144. ig 3 23 Three neighbor images of uniform target generated by a thermal camera of dominant low frequency spatial noise High frequency spatial noise generates fast spatial variations of the intensity of camera pixels This noise component creates an effect that is called Fixed Pat tern Noise If high frequency spatial noise component is strong then it should be clearly noticeable if we capture and compare several neighbor images generated by the tested camera We could notice the high frequency spatial trend present in every frame that does not depend on the frame number and the frames are almost identical Fig 3 24 50 Fig 3 24 Three neighbor images of uniform target generated by a thermal camera of dominant high frequency spatial noise It is usually considered that the frequency of 150 kHz for NTSC video or 186 kHz for PAL video systems is the border between the high and the low fre quency components It is possible to separate low frequency temporal and spatial components from the total noise using suitable low pass filter or high pass filter in a video channel Another more convenient method to separate different noise components is to capture a series of images generated by the tested thermal camera when its field of view is filled by an uniform target The captured data created 3D patio temporal array It is possible then to calculate the measures of all four noise components by carrying out the data filtering using low high
145. in darkness and poor visibility conditions Parameters of subjective image quality perceived by humans are considered as the most important parameters of surveil lance thermal cameras from the point of view of the user who wants to have the highest ranges of detection recognition and identification of targets of interest MRTD is a measure of ability to detect and recognize targets on a non uniform background MDTD is a measure of human ability to detect targets on uniform background The real targets of interest are usually located on non uniform backgrounds and MRTD is considered as the most important measure of surveillance often mil itary thermal cameras 3 1 1 MRTD The MRTD is a subjective parameter that describes ability of the imager human system for detection of low contrast details of the observed object It is a function of a minimum temperature difference between the bars of the standard 4 bar target and the background required to resolve the thermal image of the bars by an observer versus spatial frequency of the target 31 Generally MRTD is measured by determining the minimum temperature differ ence between the bars of the standard 4 bar target and the background required to resolve the thermal image of the bars by an observer for 4 bar targets of different dimensions spatial frequency The measurement results of an exemplary long range military thermal camera are shown in Fig 3 4 Fig 3 2 Image of five 4 bar targe
146. ined as the standard deviation of the output temperature dispersion caused by the system noise As it was shown in Ref 6 NGE equals NETD in case of typical thermal cameras systems of single spectral band NGE theoretically decreases with an object temperature Practically as we can see in Fig 3 33 it does not always occur due to different reasons One of them are neut ral filters used to extend camera temperature measurement range that cause signific ant suppression of the signal coming to the detectors and increase in NGE NETD value The temperature stability TS is defined as a range in which the results of meas urements carried out at different environment temperatures are located As we see in Fig 3 34 this parameter can provide crucial information to estimate ability of tested thermal camera to carry out accurate temperature measurement at real en vironmental conditions 66 20 80 180 280 380 ae a 15 z 10 eee Je a o ZA La 0 eee ns e 5 10 7 10 20 30 40 50 T C Fig 3 34 Exemplary temperature stability TS of several thermal cameras the errors of temperature measurement AT of a blackbody of the temperature T 90 C with a few thermal cameras at different temperatures of the environment T The measurement spatial resolution MSR is defined as the minimum angular dimension of the tested object when there is still no inf
147. ing particularly if imagers of several fields of view are to be tested For each FOV a separate MRTD curve should be measured both horizontal MRTD and vertical MRTD The measurement procedure can be speed up if horizontal MRTD and ver tical MRTD could be measured at the same time using the same target It can be done ifa double pattern 4 bar target is used instead of a typical single pattern 4 bar target Fig 5 2 Next the measurement procedure can shorten more if MRTD is measured only for positive contrast targets Such a solution is acceptable only is the offset is stable and was corrected to a negligible level It means that MRTD measurement can be made for positive contrast targets only for test systems of small offset and having additionally some kind of offset monitoring Generally the test team should be informed about offset temporal changes Next it is always recommended at the beginning of testing a new thermal imager to make offset measurement correct offset using available software tools and only then to carry out MRTD measurement for positive contrast targets The simplified MRTD measurement algorithm for positive contrast is presented below 1 Carry out preliminary points 1 6 of the classical algorithm 2 Set differential temperature to equal zero Regulate slowly temperature difference around zero Determine the minimal positive temperature difference when the observer recognizes the target AT Determine the minimal negative
148. ing a special high emissivity black coating This side of the target should look into the direction of the IR collimator or in the direction of the tested thermal camera The second side of the target is coated using a high reflectivity coating in order to minimize influence of the blackbody thermal radiation on the target plate temperature The targets are typically fixed to the rotary wheel or directly to the blackbody The targets manufactured using the technology described earlier are called emissive targets There is also another type of the infrared targets called reflect ive targets Both sides of such targets are covered using high reflectivity coatings When such targets are mounted to a rotary wheel in a collimator based test system then a tested imager will see through target holes a blackbody behind the active tar get and collimator enclosure reflected by the target reflective surface If a second blackbody is located inside the collimator then we can achieve a situation when both target temperature and the background temperature can be independently regu lated Fig 4 20 If the targets are properly manufactured and properly used then both two types of targets should generate the same test results However the emissive targets are preferred because collimators used in the test system optimized for reflective targets shown in Fig 4 20 are more expensive than typical collimators collimating mirror of ultra low F n
149. it is necessary to achieve relatively stable ambient temperature to enable accurate MRTD at field conditions It is typically done by using some kind of protection of the test equipment build ings with open window or door tents etc A Case of variable target test system Here a detail procedure for MRTD measurement using a variable target test system is presented 1 Write down basic test data imager type imager serial number its field of view date of the test number and names of the observers ambient temperature collimator transmittance orientation vertical or horizontal of the bar pattern 2 Fix to the rotary wheel a set of 4 bar targets keeping proper bar pattern orientation Spatial frequencies of the targets should be equal to spatial frequencies of the needed measurement points The targets should be arranged in the following order the largest target of the lowest spatial frequency becomes target no 1 the smallest target of the highest spatial frequency gets the highest number on the wheel 3 Put the tested thermal imager at the output of the collimator The imager should look into the collimator Attention the optics of the imager should be kept as close to the hole in the collimator output wall as possible 4 Turn on the imager 5 Turn on the blackbody controller 6 Wait for at least 15 minutes Give time for both the imager and the test system to reach thermal equilibrium 7 Set temperature difference equal of
150. ity model Infrared Imaging Systems Design Analysis Modeling and Testing XIV SPIE Vol 5076 p 196 207 2003 24 Schade O 1948 Electro optical parameters of television systems 1 parameters of vision and visual systems RCA Review 9 5 37 25 Sousk S O Shea P Van Hodgkin A Measurement of uncooled thermal imager noise In frared Imaging Systems Design Analysis Modeling and Testing XVI SPIE Vol 5784 301 308 2005 26 Sousk S O Shea P Van Hodgkin A Uncertainties in the minimum resolvable temperature difference measurement Infrared Imaging Systems Design Analysis Modeling and Testing XV SPIE Vol 5407 p 1 7 2004 71 27 STANAG 4349 Measurement of minimum resolvable thermal difference MRTD of thermal cameras 1995 28 Webb C Holst G Observer variables in minimum resolvable temperature difference In frared Imaging Systems Design Analysis Modeling and Testing III SPIE Vol 1689 1992 29 Wittenstein W Minimum temperature difference perceived a new approach to assess under sampled thermal imagers Opt Eng Vol 38 773 1999 30 Wittenstein W Thermal range model TRM3 Infrared Technology and Applications XXIV SPIE Vol 3436 p 413 424 1998 31 Wood L Laboratory Bench Analysis of Thermal Imaging Systems Opt Eng 15 G193 G197 1976 32 www edmundoptics com 72 4 Test equipment Some information about structure and requirements on equipment for testing thermal ima
151. keep quasi horizontal position of the thermal camera Therefore later cooled thermal cameras employ Stirling coolers or rather rarely Joule Thomson coolers The Stirling cooler is fundamentally a closed cycle compression expansion re frigerator with no valves instead it incorporates a regenerator The regenerator is a tube of porous material that has low thermal conductivity to maintain a temperat ure gradient and high heat capacity to act as an efficient heat exchanger Typical Stirling coolers operate with a sealed charge of helium which is mechanically com pressed and then allowed to expand near the dewar cold finger This expansion cools the detector and the helium is then recycled through cooler s compressor The Stirling coolers can cool the detector to the required temperature usually after 3 5 minutes from the turn on These coolers require recharging and service by the cooler manufacturer after a fixed period of time typically after about 1000 10000 hours Size and mass of these coolers depend on required cooling power The power of about 0 2 0 6 W is enough to cool a small single detector but a few times higher is needed to cool an array FPA The Joule Thomson cooler is an open cycle cooler that converts pressurized gas typically nitrogen argon CO2 to cryogenic liquid gas High pressure gas is cooled by expansion at the throttle valve flows back through the counter current heat ex changer and precools the incoming gas
152. l W Randy Weber Paul G Multispectral Thermal Imager overview Algorithms for Multispectral Hyperspectral and Ultraspectral Imagery VII Conference SPIE Vol 4381 2001 4 Breiter Rainer et al Portable sequential multicolor thermal imager based on a MCT 384x288 focal plane array Infrared Technology and Applications XX VII Conference SPIE Vol 4369 2001 5 Campana Stehen B ed The Infrared amp Electro Optical Systems Handbook Vol 5 Passive Electro Optical Systems Chapt 1 Infrared Line Scanning Systems SPIE 1993 6 Campana Stehen B ed The Infrared amp Electro Optical Systems Handbook Vol 5 Passive Electro Optical Systems Chapt 2 Forward Looking Infrared Systems SPIE 1993 7 Campana Stehen B ed The Infrared amp Electro Optical Systems Handbook Vol 5 Passive Electro Optical Systems Chapt 3 Staring Sensor Systems SPIE 1993 8 Cocle Olivier Rannou Christophe Forestier Bertrand Jougla Paul Bois Philippe F QWIP compact thermal imager Catherine XP and its evolution Infrared Technology and Applica tions XX XIII Conference SPIE Vol 6542 2007 9 Dahlberg Anders G M High resolution QWIP thermal imager for AFV upgrade Infrared Technology and Applications XXX Conference SPIE Vol 5406 2004 10 Holst Gerald C Common Sense Approach to Thermal Imaging SPIE Vol PM86 2000 11 Kaplan Herbert Practical Applications of Infrared Thermal Sensing and Imaging Equipment SPIE Vol TT75 2007
153. l derivative algorithm can be used for this voltage control However due to its low speed a combination of DMC Dynamic Matrix Control and PID al gorithm is used often in controllers of differential blackbodies All data required by these algorithms are determined on the basis of statistical analysis of measure ment results obtained during experimental trials Before the blackbody can work properly it must be calibrated Calibration ofthe blackbody is a process when the relationships between temperature of the sensors and output voltages generated by the temperature sensors are determ ined The calibration process is carried out using an external temperature probe in serted to a hole in the emitter plate close to the temperature sensors High quality certified external temperature meter of temperature resolution 1 mK and stability 2 mK or better is used during calibration 96 The relationship between temperature and voltage is typically determined at at least 10 temperature points Later the measurement data is interpolated using a high degree polynomial and the calibration function is generated in a form of a table Val ues of the calibration table are saved in EPROM memory what enables easy editing during cyclic recalibration Practically all blackbodies offered for testing thermal imagers can be at least op tionally controlled from PC via RS232 port or USB2 0 port This way of commu nication between the PC and the blackbody is more conv
154. lackbody temperature and position of rotary wheel 2 Tools for software support during MRTD MDTD measurements a Storing test conditions type of tested imagers and its serial number test date ambient temperature collimator transmittance target frequencies 107 b Calculation of target frequencies in lp mrad using user inserted data of bar width in mm c Automatic inserting current blackbody temperatures to the test table at temporal points determined by the user d Correction of influence of two test parameters ambient temperature and collimator transmittance on MRTD MDTD measurement results e Presentation of measured MRTD MDTD in graphical form or tables f Generally the software should limit requirements on the user only to make decision whether he recognize the 4 bar pattern and carry out all data analysis visualization and recording Ability to cooperate with different types of frame grabbers capable to cap ture video signals of different standards Software should enable capturing images with no compression or using compression methods that would not degrade in noticeable way quality of the captured sequence of images Test software should enable semi automatic measurement of most import ant parameters of surveillance thermal imagers noise parameters SiTF MTF distortion distortion FOV Test software should enable measurement of most important parameters of measurement thermal imagers accuracy NETD and SRF
155. lary MTF Table 3 9 Images of sine patterns generated by a thermal camera of assumed MTF function Frequency MTF Original pattern Image LP mrad 1 0 94 a 7 2 0 78 3 0 57 4 0 37 5 0 21 l 6 0 11 l 7 0 05 8 0 02 9 0 01 10 0 0 From mathematical point of view MTF is defined as MTF v C v C v 0 3 13 where C v 0 is the contrast of image of the sine pattern at near zero frequency The contrast of the image of the sine pattern is defined as Cly z ae 1 min O Iaat L min O 3 14 where Max is the maximum intensity for a pattern of spatial frequency v white peak and min is the minimum intensity for a pattern of spatial frequency v black val ley The relationship between MTF function and contrast of a sine target creates the possibility of determination of MTF by measurement of contrast of series of sine targets of different spatial frequencies Measurement of MTF using sine targets is an oldest classical MTF measure ment technique However this measurement method is time consuming because measurement of contrast must be done for a series of sine targets At the same time this method is also an expensive one particularly in case of thermal cameras Manufacturing sine targets is difficult and costly even in case of visible targets when a sine target is created by deposition of a layer of spatia
156. lator 28 is very big relaxation of requirements on so called diffraction limited collim ator Even collimator built using poorly manufactured and poorly aligned mirrors can be treated as so called diffraction limited collimator As we can see defining collimator quality using the diffraction limit as a refer ence is a risky solution because the diffraction limits can be defined in different ways Next this approach is also logically wrong We must remember than collim ator is always used as a module of a system The tested thermal imagers is a mod ule of the same system too Aperture of the tested imagers is always smaller than the aperture of the collimator This means that quality of the final image generated by the tested thermal imager is degraded by imager aberration blur imager diffrac tion blur and collimator aberration blur Therefore resolution of the collimators used in real test systems is truly limited only by aberration effects Third interferometric methods are typically used for quality checking of mir rors These methods are also sometimes used for aligning components of reflect ive off axis collimators and for characterization of overall collimator quality In opinion of the author of this book the interferometric methods are useful during aligning process but present misleading results when we want to evaluate overall collimator quality The data presented in this form wavefront accuracy of the col limator output
157. lectance than their aluminum equivalents particularly about 10 um wavelength To summarize several guidelines on coating of mirrors for IR collimators can be formulated 1 Aluminum coated primary collimating mirrors is the best option due to high reflectance and very good durability 2 Aluminum coated secondary flat mirrors are not a good choice due to possible low reflectance of this coating at a wavelength of about 10 um when working at 45 angle If the test area is limited to testing thermal cameras then gold coating of the secondary mirror is the best option if the collimator is to be used to test both thermal imagers and visible near infrared cameras then protected silver coating for secondary flat mirror is the best option due to nearly uniform high reflectance in both visible and infrared range D Mirror thermal properties There are four materials that are most often used in mirrors fabrication optical crown glass low expansion borosilicate glass LEBG synthetic fused silica and Zerodur 90 The material for mirror fabrication should be chosen on the basis of four para meters coefficient of thermal expansion cosmetic surface accuracy surface accur acy and material cost Optical crown glass often BK7 type is an old and low cost material for mir rors Crown glass has a relatively high coefficient of thermal expansion and is em ployed when thermal stability is not a critical factor We must remember that mirror surfa
158. limits determined by the accuracy para meter we can write 10 eee _ accuracy intrisic AA ae 3 22 However practically the accuracy parameter is not useful for estimation of the in trinsic uncertainty of measurement thermal cameras because the conditions in which the accuracy is measured are not typically clearly defined by manufactures The question is whether the accuracy is measured at optimal calibration condi tions when the measurement errors are the smallest or it is measured at real meas urement conditions when the errors can be many times higher Manufacturers state very rarely at what ambient temperature the accuracy was measured However typical practice is that the accuracy is measured at laborat ory conditions when ambient temperature is equal to about 23 C During real measurements the environment temperature can vary significantly within wide limits from about 20 C to about 40 C Changes of the environment temperature can have significant effect on the measurement results due to several reasons First radiation emitted by the optical elements of the camera depends dir ectly on temperature of these elements and indirectly on temperature of the environ ment Second variation of the environment temperature can cause variation of the detector temperature Third changes of the environment temperature cause direct changes of the temperature of the electronic blocks and indirect chan
159. lly variable transmit tance over a transparent glass It is technically possible to manufacture sine filters for far infrared range but it is so difficult and costly technology that MTF of thermal imagers is never measured using sine targets method Contrast transfer function CTF can be treated as a substitute of MTF that can be measured using targets that are simpler to manufacture CTF is defined in the same way as MTF with an exception that a square wave pattern is used instead of a sine wave pattern The CTF values are usually higher than the MTF values The difference between MTF and CTF is usually not great Next it is much easier to measure CTF than to measure MTF Therefore several decades ago CTF was typically used instead of MTF to evaluate thermal cameras However nowadays CTF is rarely measured because in present era of computer technology and image processing MTF function can be easily measured by capturing images of some standard targets and using mathematical apparatus of Fourier transform MTF can also be defined as the magnitude of a complex function Optical Trans fer Function OTF V MTF V exp i PTF V 3 15 58 where OTF is Optical Transfer Function MTF is Modulation Transfer Function and PTF is a function called Phase Transfer Function that represent the change in phase position as a function of spatial frequency A perfect optical system would have MTF of unity at all spatial frequencies and PTF equal to 0 a
160. luence of limited size of this object on temperature measurement results The MSR parameter is typically measured as the angular slit dimension when the slit response function SRF equals 0 99 67 STRF o O gt a mrad Fig 3 35 Slit Temperature Response Functions of two exemplary thermal cameras As we can see in Fig 3 35 there can be big differences between measurement resolution MSR of different measurement thermal cameras present on the market MSR of the first thermal camera shown in Fig 3 35 equals 3 5 mrad in situation when the MSR of the second camera equals 10 mrad The first thermal camera can be used for accurate temperature measurement of the targets of size as small as 3 5 mrad the second as small as 10 mrad Therefore is always recommended to carefully check MSR value of a measurement thermal camera before trying to make measurement of temperature of small targets 3 6 Summary Over forty parameters of thermal imagers were presented in this chapter The task of measurement of all these parameters is a very time consuming task even in case of a single thermal imager Fortunately probability that a reader of this book will be forced to measure all the parameters presented in Table 3 1 is low even if the reader is a test professional Most people actively involved in testing thermal imagers do not even know definitions and test methods of most parameters of thermal imagers shown in Table
161. ly found among the parameters of modern measurement thermal cameras It is related to the minimum angular size of the tested object for which influence of the size of the tested object on measure ment results is still negligible However this minimum size depends also on a para meter of other blocks of the thermal camera like aberration of the optical block dif fraction effects frequency bandwidth of the electrical channel and it is not possible to determine this minimum size on the basis of the IFOV only Manufactures of measurement thermal cameras present also MRTD or MDTD functions or values of MRTD MDTD functions measured for a case of large targets low spatial frequency If the thermal camera is only to be used for non contact temperature measurement on surfaces of the tested objects then MRTD and MDTD parameters are practic ally useless because it is impossible to connect these parameters with the measure ment errors of the thermal camera However MRTD and MDTD can be useful fig ure of merits if the cameras are to be used in non destructive thermal testing SAARS testing that can be treated as a part of NDTT technology Image quality is then as important as the accuracy of temperature indications 5 7 If the information provided by manufacturer is not enough and want to get more knowledge about possible performance of a measurement thermal camera then it is recommended to measure a set of four parameters 1 Minimal error ME 2
162. mager The tar get emissivity should be the same as the blackbody emissivity to avoid situation when the target is seen not due to temperature difference but due to emissivity difference Therefore target emissivity should be at least 0 97 to resemble ideal blackbody surface High reflectivity of the target side surface facing the blackbody Targets should be polished or coated to get reflectance of at least 0 9 in order to eliminate influence of blackbody temperature on target temperature As we can see in Table 4 7 the list of types of infrared targets is quite long However a typical set of targets for testing surveillance thermal imagers is rather short 1 A set of six to twelve 4 bar targets for MRTD measurement 105 2 Edge target for MTF measurements 3 Distortion target for distortion FOV measurement The set of 4 bar targets is built typically from several single targets bigger pat terns and several double 4 bar smaller patterns Single big 4 bar targets are used because simply there is no space for double patterns in the target plate if the 4 bar pattern is big The edge target is typically preferred over slit target during MTF measurement There are several advantages of edge targets but one of most important is the fact that several slit targets are needed for testing different thermal imagers in situation when a single edge target can be used for measurement of MTF of any thermal im ager The distortion tar
163. mager and the test system Next more space is required during the test 128 procedure A long corridor of 20 50 m length is typically required but in case of tests of long range thermal imagers longer distances are sometimes needed These requirements create logistical problems for some potential users of test equip ment There are however several serious advantages of the variable distance test sys tems First the measurement speed is high in case of testing portable thermal im agers or in any case when a distance change can be quick Second a big advantage of the variable distance test systems is the possibility of carrying out measurement at any spatial frequency We must remember that in case of variable target image projectors measurement or MRTD can be carried out only at spatial frequencies that are equal to spatial frequencies of the set of 4 bar targets the user posses This latter feature saves a lot of problems and money when it is necessary to test differ ent thermal imagers at different spatial frequencies Third the variable distance test systems can be optionally used at field conditions to enable testing thermal imagers located at their mechanical carriers Please note however that it is difficult to carry out accurately MRTD measurement at field conditions due to possible influence of environment conditions on measurement results It is necessary to protect the test system against wind direct sun rain etc In other words
164. mager saturates for tem peratures from the required temperature range or the screen brightness is zero also for temperatures of this range Second analysis of captured sequences if images generated by tested thermal imager can deliver information about all noise parameters NETD FPN non uni formity 1 f 3D noise model components NPSD However these parameters are calculated by test software in digital levels as frame grabbers digitize input analog video images generated by typical thermal imagers When SiTF is known then noise parameters can be converted from digital level units to more easy to under stand temperature units The following formula is used for this conversion Noise digital levels SiTF digital level mK Noise mK 5 19 Measurement principle of both the responsivity function or its simplified form SiTF is simple Tested imager is looking into the blackbody of regulated temperature Changing blackbody temperature changes brightness of the blackbody image The task of the test software is to regulate blackbody temperature and to measure brightness of the blackbody image If temperature regulation span is wide then complete re 148 sponsivity function is measured If this span is narrow and fit properly to the linear range of the responsivity function then only SiTF is measured Fig 5 14 Image of a blackbody during responsivity function measurement white mark the area where the brightness is measu
165. magers as these frame grabbers acquisition software was optimized for a task of capturing images with significant degrada tion of image quality in order to save disk memory needed for storing long video sequences There is a different situation in case of professional frame grabbers acquisition software capable to capture digital video images in the following standards Fire Wire USB 2 0 CameraLink GigE or LVDS 5 Calculation of noise parameters One of the tasks of computer programs offered by manufacturers of equipment for testing thermal imagers is to do the analysis of the video sequence and cal culate the noise parameters The offered programs differ by the numbers of noise parameters they calculate and by other features On one side there are computer programs developed only to calculate one para meter NETD and to present it in form of a number On the other side are more advanced computer programs that calculate not only a long series of noise para meters NETD FPN non uniformity 1 f noise 3D Noise components NPSD but also bad pixel number bad pixels location and provide graphical represent ation of some noise parameters The choice in this matter belongs to the user 153 of the software depending on the requirements on the test system Exemplary noise measurement results are shown in Fig 5 18 and Fig 5 19 Noise parameters depend on ambient temperature particularly in case of MWIR thermal imagers The ambient temperatu
166. mass these systems are excellent measuring tools at field depot applications when number of parameters to be measured is limited They can be packed in a large suitcase and easily transported to any location Therefore these test systems are very good tool for comparison of quality of different thermal imagers offered by different manufacturers Next it is possible to test thermal imagers from some dis tance and as a consequence the thermal imager can be tested without removing it from the mechanical carrier tank or helicopter Further on several thermal im agers can be tested at the same time using a variable distance test system Fig 4 8 Another advantage of variable distance systems is the fact that there is no limita tion on aperture of the tested cameras in case of the previously discussed group image projectors the aperture of the tested thermal camera must be at least about 10 smaller than aperture of the IR collimator Fig 4 8 Concept of group testing using variable distance test system Low offset phenomenon is another big advantage of variable target test systems In case of variable target test systems it was is possible to measure directly targets temperature because the targets are movable parts Only indirect temperature meas urement methods are possible This indirect temperature measurement method is one of the biggest source of creating the so called offset of blackbody indication of differential temperature In
167. mechanical electrical characteristics environmental characteristics and performance characteristics Measurement methods of physical and environmental characteristics of electro optical systems do not differ significantly from measurement methods of the same characteristics of other type of photonic systems The test methods can be found in numerous literature and will not be discussed in here We are to concentrate on a measurement of performance characteristics Quality of the output image is the most important criterion for evaluation of operation of electro optical imaging systems We cannot objectively judge an imaging system thermal camera NVD TV camera looking on the image ofa typical scenery Quality of the image can be bad for some people but still acceptable for others As shown in Fig 1 7 we cannot determine whether the right image is acceptable or not without proper testing Fig 1 7 Image of the same target obtained using two thermal cameras An imaging system cannot be properly evaluated on the basis of the image of a typ ical scenery but it can be properly evaluated on the basis of the image of some standard targets 4 bar targets square targets circle targets slit targets different types of resolution targets sine targets etc projected by a measuring system target projector to the tested imaging system Using proper measuring systems and hav ing a knowledge about radiometric and photometric parameters of the emi
168. ment results of an exemplary thermal camera are shown in Fig 3 8 Field of view 100 ties Se pzz ha X m 20x 13 3 Z 5 x3 3 z 5 aa S DZ 1 z z Z 0 1 Z Z 7 0 01 0 5 10 15 20 25 30 inversed spatial size 1 mrad Fig 3 8 MDTD of exemplary thermal camera 3 1 3 Evaluation of ranges of effective surveillance Detection recognition and identification ranges of a target of interest are the prime criterion for evaluation of most surveillance thermal cameras It is possible to measure directly detection recognition and identification ranges of a target of interest and to evaluate the tested thermal camera on the basis of the test results However it is a risky solution The ranges vary with observation conditions atmosphere background and it is relatively easy to manipulate with the detection recognition and identification ranges at real conditions if the observa tion conditions are not very precisely specified Next it is difficult to compare test results of different thermal cameras tested at different time periods and at different observation conditions It is possible to calculate theoretically the detection recognition and identifica tion ranges of any target man tank truck using simulation computer program based on mathematical models of thermal cameras 22 30
169. ments of laboratory set ups used for testing thermal imaging systems The function of the collimator is to generate a thermal image closely resembling the thermal scene at the target plate located at collimator focal plane 4 2 1 1 Collimator structure Reflective two mirror collimators built using an off axis parabolic collimating mirror and a smaller directional flat mirror represent a typical design of the collim ator to be used in systems for testing thermal imagers The reflective collimators like the collimator of diagram shown in Fig 4 9 dominate the market because of several reasons 1 Very high manufacturing costs of large size infrared refractive objectives needed to built refractive collimators 2 High polychromatic aberrations of the refractive collimators Non existence of polychromatic aberrations in case of reflective collimators 3 Wide spectral range of reflective collimators baffles primary collimating A5 mirror a secondary Focal plane mirror Fig 4 9 Diagram of typical reflective off axis collimator Mirrors used in reflective collimators are almost always front surface mirrors of configuration shown in Fig 4 10 and the back surface mirrors are excluded from further discussion The front surface mirrors consist of three basic elements sub strate reflective film and protective layer 81 reflected radiation Protective layer Reflective coating substrate material Fig 4 10 Fr
170. mporal stability are recommended Blackbodies of temporal stability not worse than 3 mK are usually acceptable but there are cases when temporal stability at level 1 mK is needed Non uniformity of temperature distribution on surface of the blackbody emitter should not influence the measurement results The requirements on non uniformity are the highest during measurements of subjective characteristic like MRTD The observer should not notice any non uniformity on blackbody surface during MRTD measurement Temperature difference during MRTD measurements is usually not higher than 5 C and therefore it is usually considered that non uniformity is accept able if it is below 10 mK at 5 C Tests of thermal cameras can be quite time consuming particularly MRTD measurements Therefore blackbody speed stabilization time is an important para meter because long settling time means long lasting tests There is no strict limits on settling time but it is convenient if blackbody temperature stabilize at time no longer than about 1 minute in case of small about 50 mm diameter blackbodies or 90s incase of bigger blackbodies of a size more than 100 mm Due to relative nature of measurement of most characteristics of surveillance thermal imagers accuracy of absolute temperature of the blackbody emitter is not a crucial parameter Even significant errors of differential temperature measure ment are fully eliminated or at least significantly reduced when meas
171. ms that create an image using the radiation reflected by the observed targets Imaging systems of spectral bands located in wavelengths of over 3 um belong to the first group imaging systems of spectral bands located in visible or NIR spectral range up to 1 um belong to the second group Sun stars Moon xt NVD TV camera target peo S AE R be 2 thermal imagers target d g A Fig 1 6 Principle of work of night vision devices TV cameras and thermal imaging sys tems Night vision devices and TV cameras are sensitive to a visible range and near infrared radiation up to about 1 um because the Sun the Moon and stars emit mostly in visible and near infrared spectral bands Thermal imaging systems use medium and far infrared range from about 3 um to about 15 um because targets of typical temperatures on the Earth emit mostly in this spectral band Because spectral sensitivity band of I2Ss and TV cameras at least partially overlaps the spec tral band of a human eye images generated by NVD or TV cameras are similar to the images generated by human sight Thermal imaging systems are truly infrared imaging systems that generate images that significantly differ from the images gen erated by human sight 1 4 Concept of testing Sets of technical data used in specifications of E O systems to describe their characteristics vary significantly However in general all these sets can be divided into three groups physical
172. n semi automatic testing of thermal imagers during a production process can speed up the process and reduce costs Fifthly quality of thermal imagers similarly to any other systems deteriorates with time Having proper measuring sets we can select thermal cameras that do not fulfill requirements at the end of the guarantee s period and replace them for the new ones It is possible to find additional reasons why the proper testing of thermal im agers is needed However even the arguments presented earlier show importance of testing thermal imagers for a wide community of people connected with thermal imaging technology Testing thermal imagers is quite well standardized and there is rich literature on subject of testing and evaluation thermal imagers There are several standards that regulate testing thermal imagers 28 2 3 17 valuable books that provide a lot ofuseful information in testing and evaluation thermal imagers 12 13 14 and hundreds of scientific papers on this subjects like sample Refs 5 27 26 4 7 22 However in spite of available standards valuable books and numerous literature less experienced test teams meat a lot of problems to carry out their tasks due to different reasons First testing thermal imagers is a very difficult task Extended knowledge from different areas like physics optics electronics thermal sciences precision mechanics metrology and practical experience with thermal imagers are needed to carry out
173. n an uniform bright background in case of positive contrast targets The visible targets can be also manufactured by precise printing of images of different shape on high quality paper However these two techniques are rarely used to manufacture targets for testing thermal imagers be cause it is difficult to determine and to control temperature distribution on the sur face of such targets Additionally typical glasses poorly transmit radiation over 3 micrometers Therefore infrared targets for testing thermal imagers are manufac tured using a different technology Targets for testing thermal imagers are usually manufactured by creating preci sion holes of different shapes in metal sheets When a blackbody is put behind such a target the tested thermal camera sees a target of shape determined by the holes on an uniform background The apparent temperature of this target is equal to blackbody temperature the apparent temperature of the background is equal to the temperature of the real target plate Fig 4 18 Image of a 4 bar target generated by a tested imager during MRTD measure ment 101 Fig 4 19 Photo of a multiply 4 bar target used for MRTD measurement from two sides In order to achieve high thermal uniformity on the surface of the target the tar gets are manufactured by cutting the desired holes in a thick metal sheet of high thermal conductivity typically copper alloys Next one side of the targets is coated us
174. n the average frame FPN ofan analyzed area or the whole thermal image is calculated as a standard deviation of spatial variation of intensity of different pixels within the analyzed area Calculated FPN in digital level units is converted to the FPN in temperat ure units _ FPN digital levels FPN C 6 FPN can be also expressed as a percentage of average intensity of the analyzed area or as a percentage of NETD Non uniformity l 2 U Capturing a short video sequence of thermal images generated by the tested thermal imager Averaging operation of the captured frames A group of frames is replaced by a single frame temporal noise component is eliminated or at least re duced Low pass frequency filtering operation on the average frame Non uniformity NU of the analyzed area or the whole thermal image is calculated as a standard deviation of spatial variation of intensity of differ ent pixels within the analyzed area Calculated NU in digital level units is converted to the NU in temperature units 53 _ NU digital levels SiTF 6 FPN can be also expressed as a percentage of average intensity of the ana lyzed area or as a percentage of NETD NUC 3 11 High frequency temporal noise expressed by NETD is usually a dominant noise component in cooled thermal imagers High frequency spatial noise expressed as FPN is typically a dominant noise component in non cooled staring thermal im agers L
175. n the basis of measurement results of the responsivity function 40 _ saturation Jevel 0 a 4 T T T T T temperature gt Fig 3 11 Responsivity function of a DC coupled thermal camera t turation Jevel uJ temperature difference Fig 3 12 Responsivity function of a AC coupled thermal camera or DC coupled camera with AGC automatic gain control The responsivity function is usually S shaped The signal transfer function SiTF or the responsivity is the linear part of the re sponsivity function It is calculated as tangent of the angle between linear part of the responsivity function and the temperature axis the slope of the linear part The saturation level is the upper part of the responsivity function Dynamic range is the ratio of the maximum measurable input signal and the minim um measurable input signal maximum measurable input signal Dynamic Range 3 4 minimum measurable input signal 41 The situation is rather unclear what are really these two measurable input sig nals There are at least two different definitions of the dynamic range parameters used in specifications of thermal cameras First the dynamic range is defined as a ratio of temperature difference generat ing the output signal equal to 95 or 90 or 100 of the saturation level value to the temperature resolution of the tested camera It is typically assumed that the temperatu
176. ng visible imaging systems then the collimator is required to project radiation in the spectral range from 0 4 um to 15 um Spectral range of the reflective collimators is determined by coatings of the mirrors Metallic coatings are typically used as reflective coatings in IR mirrors There are three types of most often used metallic coatings aluminum silver and gold All three types offer high reflectivity over about 95 in the spectral range of interest 1 15 um As it was discussed earlier all mentioned above coatings need some kind of dielectric overcoat that arrests the oxidation process or improves its mechanical properties Gold offers consistently very high reflectance about 99 from about 0 8 um to about 50 um Silver offers slightly lower reflectance about 97 but broader spectrum from 0 3 um to over 20 um Aluminum coatings are characterized by lower average reflectivity about 95 and a certain reflectivity drop in near infrared From the other point of view the aluminum coatings are characterized by the best durability and the lowest costs Additionally reflectance of aluminum coatings increases with a wavelength Practically there is only a slight difference in 3 15 um spectral region between aluminum mirrors or gold mirrors but only in case of collimating mirrors where a mirror surface is nearly perpendicular to the incoming beam Silver gold coated flat mirrors working at about 45 deg angle are characterized by much better ref
177. nges low medium 120 and high In case of staring thermal imagers the frequency of the smallest target the highest frequency should be about the Nyquist frequency The latter one can be calculated as 1 2 IFOV Three measurement points is the minimal number It is recommended to carry out measurements at more points Practically it means that for testing a set of dif ferent thermal imagers available on the market a set of at least twelve 4 bar targets is needed Next if the specifications of the tested imager present its MRTD values it is recommended to carry out measurements at exactly the same spatial frequencies as reported in the specifications Spatial frequency of 4 bar patterns is calculated as v lp mrad f m 2 a mm in case of variable target test systems v Ip mrad R m 2 a mm in case of variable target test systems where v is the spatial frequency of the 4 bar target f is the focal length of the col limator R is the distance between the test system target plane and the imager op tics plane and a is the width of a single bar in the 4 bar target Spatial frequencies of exemplary targets used in several variable target systems and in a variable distance system are shown in Table 5 2 Table 5 3 In the first case it was assumed that a series of 4 bar targets of a bar width variable from 0 1 mm to 4mm and variable collimator focal length from 1 m to 3 m were assumed In the second case it was assumed tha
178. nt result for n observer n 3 in case presented in Table 5 7 Fill the MRTD column 28 Correct the raw measurement MRTD results using this formula MRTD o MRTD cor T Teor 5 8 where cor T is the correction coefficient see Table 5 1 and Teo is trans mittance of the collimator Fill the MRTD o column The procedure presented above enables us to measure MRTD at one orientation The exemplary data shown in Table 5 6 and Table 5 7 refer to vertical bar ori entation that enabled to measure horizontal MRTD In order to measure vertical MRTD it is necessary to change orientation of the single 4 bar targets fixed to the rotary wheel on horizontal orientation and to repeat the earlier presented algorithm and insert the test results to Table 5 8 Attention If double 4 bar targets see Fig 5 2 are used then both horizontal MRTD and vertical MRTD can be measured at one measurement process MRTD AT Table 5 6 Exemplary test conditions No Test conditions Value 1 Imager name type number and field Marico cooled 8 12 um no 232 of view FOV 2 x1 5 2 Number of observers and names 3 F Fotell K Novak Y Russon 3 Test date November 12 2006 4 Test place Factory laboratory no 10 5 Test system type and number DT 2000 no A12 2005 6 Ambient temperature 16 C correction coefficient 0 98 7 Collimator transmittance 0 94 8 Target type and bars position Single 4 bar pattern vertical posi tion
179. nts on surveillance thermal camera using a concept of performance ranges Field of view Detection range km Good transmission o 0 2 Bad transmission o 1 Wide field of view 2 7 Narrow field of view 7 3 Recognition range km Wide field of view 1 1 0 9 Narrow field of view 3 1 2 Table 3 3 Exemplary requirements for MRTD function of a long range surveillance thermal camera Spatial frequency mrad MRTD C field of view wide about 10 narrow about 3 0 5 lt 0 1 1 lt 0 38 lt 0 1 1 5 lt 2 2 lt 0 18 3 lt 0 4 4 lt l 5 lt 6 Attention These are only exemplary MRTD values To summarize we can say that MRTD is the most important characteristic of thermal cameras from the point of view of the user who wants to have the best ranges of detection recognition and identification of targets of interest Having known MRTD functions of different thermal cameras we can calculate the ranges of detection recognition of the standard NATO target and compare their perform 37 ance only thermal cameras of almost the same field of view should be compared Therefore proper specifications of a thermal camera should specify precisely max imal values of MRTD function at a set of spatial frequencies Then the measured MRTD values must be lower than the values in the specifications if the camera is to pass the test Now let us discuss the way to
180. ocedure movement speed on sensor characteristics is another big drawback of the DMRT method There are however two real rivals of classical MRTD as the figure of merit of thermal imagers Triangle Orientation Discrimination TOD threshold as an alternative to MRTD method to characterize performance of thermal imagers 3 The TOD method proposes to use a series of triangle targets for characterization of thermal imagers Fig 3 5 Image of a triangle target during testing thermal imager using TOD method The TOD method has a number of theoretical and practical advantages over the MRTD method The method is based on an improved test pattern a well defined observer task a solid psychophysical measurement procedure and gener ates more accurate and more comparable results from different measuring teams than in case of classical MRTD It is also possible to convert TOD measurement results to MRTD characteristics However mathematical conversion algorithm is not simple Next TOD method was invented several decades after introduction ofthe MRTD concept Further on the classical MRTD test method has been already implemented in several widely disseminated military standards Therefore 34 probability that the TOD method shall be widely accepted by international com munity is rather low in spite of potential advantages of this test method Minimum Temperature Difference Perceived MTDP parameter is a new figure of merit to evaluate quality
181. odern thermal imagers can vary in time Sometimes such effects are noticeable after a dozen minutes even if the am bient temperature is stable If ambient temperature varies then stronger variation of noise parameters NETD FPN NU can be expected Such a situation most often 155 occurs in case of non cooled imagers but sometimes occurs also for thermal im agers built using cooled IR FPAs Fourth the noise parameters of thermal imager can vary significantly when the imager heats up even in case of thermal imagers built using IRFPA modules of good temporal stability Therefore the tests should be started not earlier than after at least 15 30 minutes of work of both the tested imager and the test system If this condition was not fulfilled then repeatability of test results can be bad To summarize because of inconsistent literature recommendations inherent de pendency of noise related parameters NETD FPN NU on imager setting and pos sible temporal variations the user should not expect very good agreement of meas urement noise data with data provided by manufacturer or other laboratories Agree ment at the level about 20 30 should be considered as acceptable situation 5 5 References 1 ASTM standard E 1213 2002 Standard Test Method for Minimum Resolvable Temperature Difference for Thermal Imaging Systems 2 Holst G C Infrared Imaging System Testing Vol 4 Chapt 4 in The Infrared amp Electro Optical Systems Handbook
182. of pixels in the second or the third frame are clearly darker or brighter in comparison to the first frame Fig 3 21 Images of uniform target captured at long intervals High frequency temporal noise generates fast temporal variations of the intensity of camera pixels If we refer to original interpretation of NETD in older scanning thermal cameras then NETD can be considered as a measure of this high frequency 49 temporal component of the total noise High frequency temporal component is clearly noticeable if we capture and compare several neighbor images generated by the tested thermal camera We can see in Fig 3 22 that the pixels intensity depends on a frame number in spite of the fact that they are neighbor frames and time inter vals between them are very short 1 60 s in NTSC video system or 1 50 s in PAL video systems Fig 3 22 Three neighbor images of uniform target generated by a thermal camera of dominant high frequency temporal noise Low frequency spatial noise generates slow spatial variations of the intensity of camera pixels This noise component creates an effect that is called non uniform ity If low frequency spatial noise component is significant then it is noticeable if we capture and compare several neighbor images generated by the tested thermal camera We could notice present in every frame low frequency spatial trend that does not depend on a frame number and the frames are almost identical Fig 3 23 F
183. of spectral band located within 1 3 um range MWIR spectral band and LWIR spectral band differ substantially with respect to background flux scene characteristics temperature contrast and atmospheric transmission under diverse weather conditions Factors which favor MWIR band are higher contrast superior clear weather performance favorable weather condi tions e g in most countries of Asia and Africa higher transmittance in high hu midity and higher resolution due about 3 times smaller optical diffraction Factors which favor LWIR band are better performance in fog and dust conditions winter haze typical weather conditions e g in West Europe North USA Canada higher immunity to atmospheric turbulence and reduced sensitivity to solar glints and fire flares The possibility of achieving higher signal to noise S N ratio due to the greater radiance levels in LWIR spectral range is not persuasive because the background photon fluxes are higher to the same extent and also because of readout limitations Theoretically in staring arrays charge can be integrated for full frame time but because of restrictions in the charge handling capacity of the readout cells it is much less compared to the frame time especially for LWIR detectors for which a background photon flux exceeds the useful signals by orders of magnitude To summarize in general the LWIR band is preferred for high performance thermal imaging because of its higher sensiti
184. of these imaging sys tems However in spite of available standards valuable books and numerous liter ature less experienced test teams have a lot of problems to carry out their tasks due to lack of a practical guidebook in field testing thermal imagers This book presents knowledge of the author on testing thermal imagers that was accumulated during over two decades of scientific work in the field of electro op tical technology interconnected with a series of practical projects and direct in volvement in activities of one of manufacturers of equipment for testing thermal imagers Inframet www inframet com The author hopes that this book can become a practical guide for testing thermal imagers for wide community of people interested in this fascinating technology of thermal imaging PS The Author hopes that reading this book will be much easier task than pro nouncing his family name The Author also apologizes for imperfect language of this book because English is not his native language October 2010 Krzysztof Chrzanowski iii Table of Contents 1 TARO CUCU NN cocoa dees re alse Aas cass ceccts cedceceueceen ateaceincencet ones 1 1 1 Concept of electro optical imaging SysteM cccsscssssscssscsssssscerssssesetsesseses 1 1 2 Human Sight scsscssscssssssssscssscsssssssesssssccnsssssesssssssssssssssssessesscsssscssonsesssssessesees 1 1 3 Division of electro optical imaging SySteMs cccrsrcsssesccss
185. offered on the commercial market are typically built from two blocks the radiator block and the controller block The diagram of the radiator block is shown in Fig 4 14 As it can be seen this block is built from the radiation emitter the Peltier element two temperature sensors the sensor of temperature of the radiation emitter and the sensor of temperature of the front wall the cooling radiator integrated with the fans the electronic module and the block case 93 to CTCB A D converter Inout R U converter PTR resistor Peltier power supply high emissivity radiator closing I PTR resistor aperture ____ gt I x z i I i AJ L 1 high reflectivity coating N emitter plate hole for the ex ternal probe Fig 4 14 Construction of the radiator block The radiation emitter is manufactured as a sandwich of several metal plates fixed together The plates are manufactured from material of high thermal con ductivity in order to achieve very good temperature uniformity on the surface ofthe emitter The front side of the emitter plate is painted using a special high emissivity coating This technique enables to achieve emissivity of the emitter plate equal to 0 97 0 01 Emissivity of the emitter plate can be improved up to 0 985 us ing additionally so called the micro cavities technique when an array of small mi cro c
186. ogy of manufacturing any electronic element sensitive to optical radiation Secondly an imaging system based on image intensifier tube technology is typ ically called night vision device NVD but in some international military standards is called image intensifier system 29 30 At the same time we must remember that logically thermal imagers and high sensitivity TV cameras are also night vision devices because they enable nigh vision Thirdly if we make a review of literature dealing with infrared technology then we find that there are at least eleven different terms used as synonyms of the term thermal camera thermal imager 19 thermograph 31 thermovision 23 FLIR 9 thermal imaging devices 25 infrared imaging radiometer 8 thermal viewer 1 thermal video system 32 infrared camera 18 thermal imaging device 24 If we analyze the Internet resources we can easily find even more synonyms of the term thermal camera Fourthly the most surprising thing is that actually even the term infrared radi ation or division of IR radiation is not standardized There was presented in the In ternational Lighting Vocabulary considered nowadays as the international primary authority on terminology in radiometry the range of optical radiation a proposal of division of optical radiation see Table 1 2 but not as compulsory division but only as a recommended division Additionally in case of visible radiation due to
187. on some limitations of spatial resolution of the human eye exist it means that there are some limits how small can be the targets that the eye can still perceive This limit depends on the target contrast and illumination conditions Generally it is con sidered that at bright moon night illumination 3x10 cd m the resolution is about 1 5 mrad but in bright day conditions illumination 310 cd m the resolu tion is 0 35 mrad 21 This means that when the distance to the targets is 25 cm then human cannot recognize as separate two slits of a width smaller than 0 37 mm for weak illumination and 0 09 mm for strong illumination There are generally two types of vision The first type photopic vision day light vision when the cones are active as light sensors The second type scotopic vision nightlight vision when the rods are active as light sensors The cones are active at illumination levels over 0 03 cd m The upper limit for proper work of the cones is about 300 000 cd m The rods are activated by illumination conditions be low 0 03 cd m After sufficient dark adaptation time even 30 minutes is some times needed the rods can enable us seeing even at 0 03 cd m However quality of an image generated by the rods in night conditions is much lower than quality of an image generated by the cones in daylight conditions Next the rods are not sens itive to a wavelength of the incoming light and therefore cannot perceive
188. on cooled thermal cameras is inferior to image quality offered by cooled cameras but is good enough to be used in many short and medium range ap plications Due to a 2 4 times lower price than equivalent cooled systems the num ber of non cooled thermal cameras is growing rapidly in both military and commer cial applications Parameters of thermal cameras from the same generation can vary significantly Therefore it is not possible to form a single table enabling accurate comparison of parameters of thermal cameras from different generations Table 2 1 was created on the basis of a review of the parameters of different observation thermal cameras offered during the last 30 years but should be treated as an estimation of the sophist icated situation on the market Table 2 1 Typical parameters of thermal cameras resolution resolution kg NETD K Gen 60 120 pixels CMT US 0 2 250x190 liquid ni 1 common modules trogen 8 14 pixels CMT SPRITE Joule US UK common modules Thomson 0 1 2 0 05 Gen 94x4 pixels CMT Ophelios 640x288 Stirling 288x4 CMT Synergy Cath Joule erine Sophie Iris 320x240 bolometric Elvir 320x240 Gen 320x240 HgCdTe MWIR 320x240 3 Opal Spike Matiz 320x240 QWIP LWIR 640x480 Thermovision 2000 Cath micro erine QWIP scanning 640x512 HgCdTe MWIR High Definition POD 320x240 ferroelectric Lion 23 Fig 2 5 Exemplary non cooled Gen 3 thermal camera ELVIR courtesy of
189. on of infrared radiation used in this book Wavelength range near infrared NIR 0 78 um 1 um short wave infrared SWIR 1 um 3 um mid wave infrared MWIR long wave infrared LWIR very long wave infrared VLWIR 15 um 1000 um It is possible to present many more examples of a certain chaos in the termino logy related to electro optical systems The aim of the author of this book is to use the precisely defined and strictly kept terminology in all book chapters However the reader must be aware that in a situation of general chaos in terminology re lated to electro optical imaging systems it is difficult to achieve this aim and there can be some inconsistency in terminology used in the book 1 6 Basic metrological terms Testing electro optical imaging systems practically means to measure the para meters of these systems A measurement is a non accurate operation Measurement results always differ from the true value of the measured quantity Equality of the measurement result and the true value of the measured quantity is an exception al incident and we do not know when such an incident occurs It is recommended for the test teams to be familiar with some basic metrological terms Accuracy of a measurement result can be only estimated It can be done using classical error theory or modern uncertainty theory Classical error theory proposes so called limit error as a measure of measurement accuracy Models that can be used for determina
190. on one the imager is looking into the collimator Version two the imager is looking into a large blackbody located at some distance within imager focus range Version three the blackbody is loc ated just behind the imager optics 3 Marking area of interest The blackbody that simulates the uniform target typically fills entire field of view of tested imager However sometimes some parts of video image gen erated by the tested imager are not active black frame around the true active im age see Fig 5 17 This effect is usually met in thermal imagers using IRFPA modules optimized for resolution of NTSC standard but generating output video image in PAL standard If such a case occurs then the user must mark area of the active part of the image to be analysed If not measurement results of spatial noise parameters can be significantly distorted 4 Capturing of a sequence of images Capturing of a video sequence of images of the uniform target is crucial part of procedure to measure noise parameters The capturing should be done using high quality analog frame grabbers case of imagers generating signal in PAL NTSC standard that do not distort captured images The same require ments are also valid for the test software The compressing algorithms of cap tured frames should not distort the output images in noticeable way Most typic al frame grabbers and video acquisition software are not useful for the task of capturing images from thermal i
191. on undersampled imagers Infrared Imaging Sys tems Design Analysis Modeling and Testing XIV SPIE Vol 5076 2003 EA 3 02 The Expression of Uncertainty in Quantitative Testing previously EAL G23 European Cooperation for Accreditation 1996 Forrai David Bertke Stephen Fischer Robert Back Thomas Test techniques for high perform ance thermal imaging system characterization Infrared Imaging Systems Design Analysis Modeling and Testing XVII SPIE Vol 6207 2006 Griffith B et al Buildings research using infrared imaging radiometers with laboratory thermal chambers Proc SPIE Vol 3700 p 502 513 Thermosense XXI Dennis H LeMieux John R Snell Eds Publication Date 3 1999 Gruben J et al Scene based algorithm for improved FLIR performance Proc SPIE Vol 4030 p 184 195 Infrared Imaging Systems Design Analysis Modeling and Testing XI 2000 Guide to the expression of uncertainty in measurement International Organisation for Standarisation International Electrotechnical Commission International Organisation of Legal Metrology International Bureau of Weights and Measures TAG 4 WG 3 1993 Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results NIST Technical Note 1297 1994 Holst G C Electro Optical Imaging System Performance SPIE 2005 Holst G C Infrared Imaging System Testing Vol 4 Chapt 4 in The Infrared amp Electro Optic al Systems Handbook Mich
192. ont surface mirror 4 2 1 2 Requirements The task of the collimator is to project with negligible distortion an image of temperature distribution on the target plate located at the collimator focal plane into direction of the tested thermal imager The condition on negligible distortion can be fulfilled only when certain requirements on four parameters of IR collimat ors 1 Resolution 2 Aperture 3 Spectral range transmittance 4 Thermal properties are fulfilled A Resolution According to a popular myth the function of the collimator is to generate a paral lel ray beam in direction of the tested imager Practically the collimator does not generate a single parallel ray beam it generates an infinite number of parallel ray beams in different directions The true task of the collimator is to generate a thermal image closely resembling the temperature distribution at the target plate located at collimator focal plane In its ultimate form an ideal collimator would be capable of generating a radiation pattern that exactly reproduces the real temperat ure distribution of the target plate However such quality is unattainable Instead a practical design condition should be adopted based on the requirement that the collimator spatial resolution should match the spatial resolution capabilities of the tested thermal imager We should remember that collimators of too low qual ity can become a source of significant measurement errors collimat
193. ors of too high quality can unnecessarily increase cost of the test system Manufacturers of IR collimators use different ways to characterize performance of these optical instruments and there is a confusion in area of evaluation methods First accuracy of manufacturing of the collimating mirror is often presented as a collimator parameter 35 Values of proclaimed surface accuracy of mirrors used to built IR collimators vary in quite wide range from about 2 2 to 4 12 However manufacturing accuracy of the mirrors is not very useful as criterion of overall qual ity of collimators for testing thermal imagers Generally perfect mirrors do not necessary mean that the collimator generates a perfect image Very precise alignment of these two collimator mirrors is required 82 to obtain the maximal theoretically possible performance Next precision zero thermal expansion optical and mechanical elements must be used in collimator design Further on even small imperfections of mirrors coating or simply dust can degrade quality of images projected by the collimator Finally we must remember that the off axis parabolic mirrors are aberration free only at their focus point not for entire area of the target located at the focal plane Practically this means that information about mirror manufacturing accuracy does not give precision informa tion about overall collimator performance Practically increasing manufacturing accuracy of the collimating mi
194. oseness of the agreement between the result of a measurement and true value of the measurand where the measurand is a specific quantity subject to measurement Error of measurement VIM 3 10 result of a measurement minus the value of the measurand Random error VIM 3 13 result of a measurement minus the mean that would result from an infinite number of measurements of the same measurand carried out under repeatability conditions Comment By means of statistical analysis it is possible to estimate the random er ror Systematic error VIM 3 14 mean that would result from an infinite number of measurements of the same measurand carried out under repeatability conditions minus the value of the measurand Comment The systematic error equals to error minus random error Similarly to earlier defined terms measurand and error it cannot be fully known it can be only estimated Uncertainty of measurement VIM 3 9 a parameter associated with the result of a measurement that characterizes the dispersion of the values that could be reas onably attributed to the measurand the parameter mentioned above is usually a standard deviation or a given multiple of it The term limit error is not included into the VIM However on the basis of ana lysis of the Ref 10 it can be defined as presented below Limit error a range around the result of the measurement in which the true value of the measured quantity is located with hig
195. ossible to shorten measurement time bar target pattern vertical of MRTD because both horizontal 4 bar and horizontal 4 patter and vertical 4 bar patterns are seen bar pattern at the same time Multiple multiple 4 bar MRTD cost effective solution for MRTD 4 bar patterns of different measurements when test are to be carried targets size cut in a single out at 2 3 frequencies A single multiple metal sheet 4 bar target with several 4 bar patterns can replace a series of 4 bar targets with a single pattern 10 multiple multiple circular MDTD cost effective solution for MDTD pinhole patterns of different measurements A single multiple pinhole diameter cut in a single metal sheet target with several pinhole patterns can replace a series of pinhole targets with a single pattern 104 11 Abingdon single Abingdon testing targets are used to evaluate the cross cross pattern cut in tracking effectiveness of image processing target metal sheet systems algorithms in presence of noise 12 distortion set of narrow lines distortion to evaluate linear and angular target creating multiple displacements due to distortion effect square pattern 13 gray set of small squares response can speed up measurement of response target of different function parameters transmittance 14 silhouette silhouette pattern evaluation targets are used for evaluation of target resembling real of surveillance ranges of
196. ot use a collimator dur ing the tests then the distance between the target of the test system and the tested imager must be long enough to have the measuring system within the focusing range of the tested imager This means that the test system must be always located at distances longer than the minimal focus distance of the tested imager In case of surveillance thermal imagers the latter value vary from about 5 m to about 50 m This means that the tests cannot be carried in small rooms but should be carried out in the field or at indoor conditions using long corridors The size of the targets simulated by variable distance test systems must be big enough to enable simulation of big objects low spatial frequency 4 bar targets when the distance test system tested imager is equal to minimal focus distance of the tested imager Practically this means that due to long distance between tested imager and the test system from about 5 m to about 50 m large area blackbodies and large IR targets sizes over 200 mm in most cases are needed for variable dis tance test systems Therefore both blackbodies and targets used for variable dis tance test systems are several times bigger than blackbodies targets needed for typ ical variable target test systems sizes between about 50 mm to 100 mm 75 distance Tested imager Fig 4 5 Photo of the LAFT mobile variable distance test system courtesy of Inframet www inframet com
197. other two types of test systems variable distance systems or variable target distance systems for testing modern thermal imagers However the clas sical variable target test system is typically the preferred choice It can be estimated that over 90 of systems used all over the world for testing thermal imagers are variable test systems At the same type the other two types of test systems can be treated as simplified versions of the variable target test system Therefore from now we will concentrate only on variable target test systems and will discuss in de tail only blocks of these test systems The variable target test systems are built from the following blocks Collimator Blackbody Rotary wheel Set of targets PC Frame grabber Test software A set of the standard targets metal sheets with holes is fixed to the rotary wheel placed at the focal plane of the collimator One of the targets is within the field NYDN BRWN Ke 80 of view of the collimator we can call it the active target The differential black body is located close behind this target The temperature distribution on the target surface and the blackbody is projected by the collimator to the tested thermal cam era and image of the active target is generated by the tested imager Next the image is evaluated by an observer or the image captured and analyzed with help of spe cialized hardware software module 4 2 1 Collimator Collimators are typical ele
198. ound frame removal 61 4 Calculate PVF using the formula 3 21 Both the measurement spatial resolution MSR and the imaging spatial resolu tion ISR are Slit Response Function related parameters The measurement spatial resolution MSR is defined as angular slit dimension for which Slit Response Func tion SRF of the tested thermal camera is equal to 0 99 The imaging spatial resolu tion ISR is defined as angular slit dimension for which the Slit Response Function SRF of the tested thermal camera is equal to 0 5 ISR IFOV MSR angle mrad Fig 3 31 Graphical definitions of MSR and ISR The imaging spatial resolution ISR is typically found among the parameters of older scanning thermal cameras ISR is a good measure of camera ability to cre ate a thermal image of small targets However this parameter does not give inform ation whether the size of the tested object is high enough to assure negligible influ ence of this size on measurement results during non contact temperature measure ment The information is provided by the measurement spatial resolution MSR When angular size of the tested object is higher than the measurement spatial resol ution then we can assume that influence of the size of this object on the temperat ure measurement results is negligible We can say in other words that if the angular size of the tested objects varies but is always higher than MSR then the output temperature will be the same Howev
199. ow frequency spatial noise non uniformity is typically much lower in cooled thermal imagers than in non cooled thermal imagers Examples of possible measurement results are shown in Table 3 7 Please note however that these are only the examples and within each technology test results can vary significantly Next measurements are usually done immediately after the imager internal calibra tion and therefore the non uniformity measurement results are lower than during real work Table 3 7 Examples of measurement results of noise components of different thermal imagers Imager type NETD C 1 f C FPN C NU C cooled scan 0 1 0 04 0 07 0 08 ning cooled array 0 05 0 05 0 03 0 1 non cooled ar 0 12 0 09 0 15 0 3 ray 3 4 Image resolution parameters Image resolution parameters carry out the information about imager ability to perceive small details of high contrast images There is a lot of confusions in literat ure in this area because many parameters are used to expressed this ability Gener ally parameters that represent thermal imagers ability to perceive small details resolution can be in four groups 1 Parameters based on basic specifications of the IR FPA module number of detectors pixel dimensions used in thermal imager 2 MTF modulation transfer function and derivative parameters 3 Parameters based on imager response to point sources or slit sources 4 Parameters base
200. peed up by using several available computer programs optimized for the task of calculation of performance ranges of thermal cameras 3 2 Response parameters Response parameters give us information about system response to variable temperature targets or to the variable size targets There are three commonly used response parameters of thermal cameras 1 Responsivity function 2 ATF Aperiodic Transfer Function 3 SRF Slit Response Function Responsivity function is the system response to a large target of variable temper ature It provides information on gain linearity dynamic range and saturation level The signal transfer function SiTF is the linear part of the responsivity func tion Aperiodic transfer function ATF is defined as a normalized dependence of sys tem response to a variable size square circular target It provides information on system ability to detect small targets Slit response function SRF is defined as a normalized dependence of system response to a variable size slit target It provides information on system ability to detect long narrow targets 3 2 1 Responsivity function Responsivity function is a function of an output signal screen luminance or elec trical signal versus target temperature absolute or relative in case of a large con stant size target Fig 3 11 Fig 3 12 It can be characterized by three digital para meters SiTF saturation level and dynamic range that are determined o
201. perature range below 45 minutes Table 4 9 Requirements on blackbody Parameter Requirement Blackbody emitter size at least 10 bigger than diameter of optics of tested imager at least four times bigger than size of IR FPA sensor Work range ambient temperature 30 C to 60 C Differential temperature range 10 C 40 C Absolute temperature range 40 C to 80 C 111 As we can see in Table 4 9 the requirements on blackbody to be used for cor rection of spatial noise of thermal imagers differ significantly from requirements on blackbodies to be used in typical test systems that were discussed in Section 4 2 2 We must remember that generally typical test systems are designed to work at laboratory conditions in a situation when the tests of spatial noise are done in ex treme temperature conditions Therefore there are big differences in requirements on work temperature and on absolute temperature range Next there are also some differences in requirements on size of the blackbody emitter In case of typical test systems a blackbody of emitter size equal to 50x50 mm can be considered as suffi cient because the holes in the targets are typically smaller than about 40 mm The blackbodies that are to be used in temperature chambers they must be bigger than diameter of the optics of tested imager Optics of a diameter of about 100 mm is commonly met in many non thermal imagers Therefo
202. presented and discussed earlier There were also presented requirements on blocks of such test systems However any potential user or buyer of equipment for testing thermal imagers should be reminded that total value of such a test system is much more than a sum of system blocks One weak block reduces greatly value of a test system Thus writing detail requirements on all blocks of a test system optimized for specific ap plications is an expert task It is quite easy to make an error on details as there are many parameters that are inter dependent Now we will present several recommendations for potential buyers users of systems for testing thermal imagers If these recommendations not implemented then there is a high probability that the buyer will get a costly test system that is not optimal for his applications First it is recommended for potential buyers of test systems to concentrate more on requirements on the test system than on requirements on blocks of the system The best way to write safe for the user buyer requirements on test system is to write a list of required test capabilities of the system understood as a list of paramet ers to be measured or list of other system functions It is a common error found in many tender requirements that detail sometimes not needed requirements on col limator blackbody rotary wheel are presented but a detail list of parameters that are to be measured is not presented Second equipment for tes
203. quency targets are used b set brightness level to average position c set edge improvement electronic boost off when MRTD is measured at low frequency range d set edge improvement electronic boost on when MRTD is measured at medium high frequency range These guidelines should not be treated as must to do conditions There are im agers when at different settings better test results can be get 125 There are many thermal imagers with built in display For such imagers two display configurations are possible 1 test are carried out using original imager dis play 2 tests are carried out using a special high quality monitor The first approach is recommended as only then we can get information about real capability of tested thermal imagers 5 1 7 Static dynamic images Classical MRTD definition refers to static targets and MRTD is generally a measure of imager ability to detect recognize and identify static targets However during real work of thermal imagers the image of static targets becomes slightly dynamic due to imager movement human hands vibration when portable imager is kept movement or vibration of a tank equipped with a thermal imager etc Next the head of the observer is rarely fully static during real work too Therefore it looks logically that the observers should be allowed to move slightly both the im agers and their heads during MRTD measurement to simulate real work conditions It can surprise bu
204. r approach If we look into technical data offered by manufacturers of most thermal cameras we find a parameter called thermal sensitivity thermal resolution temperature resolution or NETD that provide information about noise of the thermal camera The parameter mentioned above has different names but usually means noise equi valent temperature difference NETD The problem is that different definitions and different measurement techniques of NETD are used in literature According to its classical definition NETD is defined as the blackbody temper ature difference between a target and its background required to produce a peak sig nal to ems noise ratio of unity at a suitable point in the output electrical channel This definition was developed at the time when all thermal cameras were the scan ning thermal cameras Although the definition does not state it clearly NETD is a metric of only high frequency temporal noise along the video line Fig 3 18 Low frequency temporal changes are corrected Fig 3 19 before the NETD measure ment Next NETD typically gives no information about the spatial noise between different video lines Fig 3 18 Signal profile of a video line Fig 3 19 Signal profile of a video line after low frequency trends correction NETD can be calculated as NETD V SiTF 3 5 47 where v is ems value of noise in the signal line SiTF is the Signal Transfer Function of the tested t
205. rature difference is kept constant but spatial frequency varies Practic ally this means that the blackbody controller keeps constant temperature difference but the distance between the test system and the tested imager is slowly increased spatial frequency is increased until the observer finds that this is the highest dis tance when he still can recognize the 4 bar pattern The procedure is repeated by several observers and then new temperature difference is set and measurement of anew MRTD point is carried out 127 100 MIRTD K 0 1 0 01 4 5 6 3 IIll Mii 1 l ll Spatial frequency V 1 mrad Fig 5 5 MRTD measurement results of three observers using the constant spatial fre quency and variable temperature measurement method 100 MRID K 0 1 0 01 4 5 6 3 IIll mi 4 l ll Spatial frequency V 1 mrad Fig 5 6 MRTD measurement results of three observers using the constant temperature difference and variable spatial frequency method Both methods should deliver the same final MRTD curve The advantage of the first method is semi automatic control of the temperature difference and the spatial frequency target number using the test software Next the vari able target test systems that use this method require little space to carry out meas urement and the tests can be carried out even in a small room The variable distance test systems require usually manual change of the distance between the i
206. rature range temperature range Stability 20 mK non cooled 2 mK typical situation imagers 1 mK extreme situation Absolute uncertainty 0 2 or 0 2 C T 25 C 2 15 mK blackbody exceeding real requirements 4 2 3 Rotary wheel The task of rotary wheel is to enable speedy exchange of the target to be projec ted by the collimator or to be observed directly by the tested thermal imager This task can be also done by horizontal or vertical sliders with a set of different targets However due to simplicity the rotary wheels are used typically to exchange the targets in systems for testing thermal imagers Two basic types of rotary wheels are available on the market manual rotary wheels or motorized rotary wheels A touch of human hand can change slightly temperature of the wheel In situation when modern thermal imagers are becoming extremely sensitive any such temperature variation can cause some measurement errors Therefore motorized rotary wheels are recommended when testing high sensitivity thermal imagers Next air flow can cause some variations of target temperature This effect can be reduced if the wheel with targets is put inside a closed enclosure that eliminates ex change of air inside outside the wheel block 99 Number of holes for the targets can vary However there are usually 6 to 12 holes for targets in the wheel The wheels are usually covered with a black high emissivity coating The requirements on co
207. re at which the measurement is carried out can differ significantly from a typical laboratory temperature about 20 C In such a case the measurement results should be corrected Most of computer pro grams used to support testing of thermal imagers can do this task Storing noise parameters Storing measurement results is a simple operation for modern software How ever practically it is extremely important to enable storing not only the meas urement results in form of several numbers but storing measurement results to gether with all parameters that describe measurement conditions gain level im age enhancement settings of the thermal imager ambient temperature test date test place test team etc Such data is very useful if comparison of test results of different imagers is to be made Fig 5 17 Thermal image with non active parts __ Noise measurement lel Noise measurement results NETD FPN NU Total NETD Total Absolute mK 36 1 110 3 487 9 501 5 116 1 MBview Store SiTE 130 di K Fig 5 18 Table with three basic noise parameters NETD FPN non uniformity 154 Psp noise analysis TF Sum Sw Sm Sw Sv S Sr S 1 10CSTDey 14 83 ld 4D Dl Random 3D noise Fig 5 19 Window of TAS T program presenting calculation results of VHT components of 3D noise model 5 4 3 Interpretation of measurement results There are quite many reasons why interpretation of nois
208. re blackbodies of a size at least about 125 mm can be considered as acceptable for spatial noise tests of major ity of thermal imagers Theoretically it is possible to use one big blackbody both for testing calibration spatial noise correction of surveillance imagers and for typical tests of thermal im agers However such compromise solution is not convenient for the user due to ne cessity of manual moving heavy blackbody test speed is reduced and finally black body reliability can deteriorate too Therefore it is recommended to use two spe cialized blackbodies a smaller blackbody as a module of a system for testing thermal imagers and bigger blackbody for spatial noise correction 4 3 2 Calibration of measurement thermal imagers Calibration of measurement thermal imagers is a process where relationship between temperature of a blackbody and output signal generated by tested imager is determined for different settings of the imager The calibration principle is simple the imager is looking into a blackbody located in its field of view as shown in Fig 4 25 The environmental requirements on measurement thermal imagers are not as harsh as such requirements on surveillance imagers Next due to some design differences the influence of ambient temperature on image generated by measurement imagers is not as significant as in case of surveillance imagers Therefore it is recommended to carry calibration of measurement thermal imagers at sev
209. re resolution is equal to NETD of the tested camera AT NETD where ts is temperature difference generating the signal that equals 95 90 or 100 depending on literature source of the saturation level Second the dynamic range is defined as a ratio of the upper value to the lower value of the temperature difference when the deviation between the response func tion RF A7 and its linear approximation is within specifications AT AT dynamic range 1 dynamic range 2 _ saturation level RE Sout gt 0 T T T T T T T T T I AT AT temperature Fig 3 13 Determination of the dynamic range using the linearity range concept 3 2 2 Aperiodic Transfer Function Thermal camera can detect small targets of angular size smaller than its instant aneous field of view IFOV The latter parameters is calculated as a detector single pixel angular substance The output signal generated by such a small size target depends on the target area For an ideal thermal camera the signal is propor tional to the target area when the target area is smaller than IFOV the signal does not depend on the target area when the target area is bigger than IFOV Fig 3 14 42 Aperiodic transfer function ATF is the dependence of a normalized function of the output signal voltage current digital on a variable size circular square target The difference between the ideal and real ATF is caused by the image blur
210. real targets targets surveillan ce ranges A process of cutting holes in metal sheets necessary to manufacture IR targets looks simple However practically it is quite difficult to manufacture proper IR tar gets that can fulfill presented below requirements l High thermal uniformity of temperature distribution on target surface The uniformity of the temperature distribution on the target plates should be a few times better than MRTD or MDTD values obtained with these tar gets In order to fulfill this requirement the targets should be manufactured from material of high thermal conductivity Copper or copper alloys are acceptable but steel sheets should be avoided Next the metal sheets cannot be too thin as thin sheets are characterized by low thermal conductance even if manufactured from proper material It seems that 0 3 0 5 mm can be considered as minimal thickness of metal sheets for target manufactur ing even in case of copper sheets High accuracy of pattern manufacturing If there are significant differences between area of the bars in 4 bar targets then accuracy of MRTD meas urement is reduced because spatial frequency of such a target is determined with some error In general the following tolerances are recommended 2 for patterns of minimal dimension over 1 mm and 4 for patterns be low 1 mm but over 0 3 mm and 8 for pattern below 0 3 mm High emissivity of the target side surface facing the tested i
211. red Typical measurement algorithm of the response function is presented below 1 Tested imager is looking via collimator into the blackbody of regulated temperature Changing blackbody temperature changes brightness of the blackbody image Test software acquires images generated by the tested imager 2 User marks an area of the blackbody image to be analysed 3 User determines the temperature regulation range of the blackbody and temperature regulation step to be used during measurement of the respons ivity function 4 The test software carried out measurement of output brightness of the blackbody image for different values of blackbody temperature 5 The responsivity function is presented in a graphical form and table form 149 Temperature d a 15 Fig 5 15 A table with exemplary data of measurement of responsivity function measurements for low gain settings 5 4 Noise parameters 5 4 1 Measurement concept The list of noise parameters is quite long NETD FPN non uniformity 1 f noise 3D noise model components NPSD However measurement of all noise parameters is based on the same concept of capturing a sequence of images generated by tested imager when viewing an uniform target Such a sequence of images carried out information about temporal noise spatial noise and noise frequency spectrum X Fig 5 16 Image of an uniform target used for measurement of noise parameters 150
212. rizontal MRTD and vertical MRTD People usually prefer to get a single MRTD curve instead of MRTD curves The problem is that we will get slightly dif ferent MRTD curves depending on averaging method we choose to average tem peratures or to average spatial frequencies to use algebraic averaging or geometric averaging Literature proposals are inconsistent in this matter The author typically uses frequency averaging using algebraic mean method because this way can be used for all measurement points Var Vit V 2 5 9 where v is the spatial frequency of horizontal MRTD curve at AT level v is spatial frequency of vertical MRTD curve at AT level Va is average spatial frequency of MRTD curve at AT level The algorithm to calculate average MRTD curve is based on three steps 1 Determine functions that approximate interpolate measured MRTD and MRTD data Two functions are obtained MRTD v and MRTD v 2 Calculate v and v at several levels of temperature difference AT MRTD and MRTD v 3 Use formula 5 9 to calculate Va at several levels of temperature difference AT After carrying out these three step algorithm the average MRTD can be calculated as in exemplary case presented in Fig 5 7 Please note that other options of calculation of average MRTD are also possible as long as it is clearly stated in the test report what method was used 133 The presented earlier in detail classical measurement procedure is time consum
213. rror does not always increases the collimator per formance but always increases manufacturing costs of the collimator Second collimator manufacturers often claim that the collimator is diffraction limited 25 27 34 This claim suggests that the collimator is perfect Practically such a claim can be very misleading Let us look on the diffraction limited target frequency values for typical collim ators produced by a calculator available at a website of one of manufacturers of equipment for testing thermal imagers 28 U S Metric Optical System Diameter inch mm Clear Aperture or Lens 5 9 150 Diffraction Limited Resolution Visible Infrared nm um Wavelength 550 3 5 8 12 Cycles mrad 55 9 10 2 6 1 3 8 2 6 Clear Form Go back to list of calculators Fig 4 11 Window of a calculator of diffraction limits The diffraction limits of several collimators calculated using this website calcu lator are shown in Table 4 3 Table 4 3 Diffraction limited target frequency values in lp mrad for collimator of different optical apertures Aperture 100 mm 150 mm 200 mm 250 300 mm Wavelength 5um 4 1 6 1 8 2 10 2 12 3 12 um 1 7 2 6 3 4 4 3 5 1 http www electro optical com eoi_page asp h Diffraction 20Limits 83 As we can see in Table 4 3 values of diffraction limited target frequencies of typical collimators are low actually ver
214. rt of MRTD measurement 8 5 1 11 Interpretation of MRTD measurement results MRTD tests generate two functions horizontal MRTD v and horizontal MRTD v that can be graphically represented as two separate curves MRTD re quirements usually do not refer directly to horizontal MRTD or vertical MRTD but to MRTD in general As it was mentioned in Section 5 1 9 there is no consensus in specialist literat ure what averaging method should be used algebraic mean or geometric mean temperature averaging or frequency averaging Literature proposals are inconsist ent in this matter The author typically uses frequency averaging using algebraic sum because this method can be used for all measurement points However other options are also possible as long as it is clearly stated in the report what methods was used Even when MRTD test results are presented in form of a single curve it is often not easy to decide whether the tested imager fulfills the MRTD requirements or not Now let us assume that after measurements we get the average MRTD curve shown in Fig 5 7 Next let us consider several versions of requirements on im ager MRTD shown in Table 5 12 and find out whether the tested thermal imager fulfills these requirements 141 Table 5 12 Exemplary requirements on MRTD of a thermal imager frequency Ip mrad MRTD requirements Version A Version B Version C Version D 0 5 lt 0 12 C lt 0 02 C lt 0 1 C
215. s Noise parameters Image resolution parameters Geometric parameters Accuracy parameters Spectral parameters Operation parameters CO SON Se a Parameters of thermal imagers Subjective quality Image Response i resolution Noise Geometric Accuracy Spectral Fig 3 1 Division of parameters of thermal imagers Subjective image quality parameters give information about ability of the sys tem thermal camera human observer to detect recognize and identify targets at different scenarios Response parameters give information about response of the thermal camera to variable size or variable temperature targets 29 Noise parameters about noise that limits camera sensitivity detecting low con trast targets Image resolution parameters carry out information about camera ability to per ceive small details of high contrast images Geometric parameters give information about geometrical relations between the target and its image Accuracy parameters give information about accuracy of non contact temperat ure measurement using thermal cameras Operational parameters provide information about position of the observed tar get or position of human eye necessary for proper operation of the camera Finally spectral parameters provide information about camera responsivity versus wavelength Subjective image quality
216. s offset AT AT_ 2 5 13 Attention AT or AT can be positive or negative numbers Insert the offset value to blackbody software memory please refer to blackbody operational manual Blackbody indications are corrected and offset influence is eliminated Regulate temperature difference of the test system to get value equal to 12 mK If the 4 bar pattern cannot be resolved then increase the temperat ure difference to 25 mK or more 50 mK 100 mK Increase gradually a distance between test system and the tested imager to find the longest distance when you can still recognize the 4 bar pattern two patterns can be used big or small Convert the value of the distance in meters to spatial frequency in lp mrad using Table 5 3 Insert the measured value of the spatial frequency to Table 5 11 Increase temperature difference to 25 mK and repeat points 7 9 Increase temperature difference to 50 mK and repeat points 7 9 Increase temperature difference to 100 mK and repeat points 7 9 Increase temperature difference to 200 mK and repeat points 7 9 Increase temperature difference to 400 mK and repeat points 7 9 Increase temperature difference to 800 mK and repeat points 7 9 Increase temperature difference to 1600 mK and repeat points 7 9 Repeat points 6 17 by the second observer Repeat points 6 17 by the third observer Calculate average spatial frequency of the 4 bar target that can be resolved using this formula Va Vi
217. s are very important main test capabilities number of parameters that can be measured or other system capabilities accepted standards of output electronic images from thermal imagers boresighting capabilit ies etc There are also some technical details like dimensions and mass of sys tems blocks of no real importance for typical applications Other parameters like blackbody accuracy blackbody temperature range are important when testing commercial thermal imagers for non contact temperature measurement but not crit ical when testing surveillance thermal imagers Further on if the tests are to be done at laboratory conditions then wide working ambient temperature range offered by one manufacturer is not a real advantage over a system from another manufacturer capable to work only at laboratory conditions We have a totally dif ferent situation if the blackbody from the test system is to be used also in temperat ure chambers for calibration applications Therefore it is always recommended to think carefully about requirements on the test system and limit these requirements only to truly needed level This recommendation is important as equipment charac terized by some non typical parameters is often much more costly than typical test system To summarize both cost effective buying and effective use of systems for test ing thermal imagers are difficult tasks Well educated properly trained test team is needed in both these tasks 163 A
218. scsscssesssssrssesees 6 1 4 Concept Of testing ccccsccscssccrssssccrssssserssssscssssssssscssssscsssssscssssssesscssesssssessesees 8 1 5 Terminology scccccsscssccsscessssscessssscsssssscesssssssssssssssssssssscsssssssessssssssssscsssssssoesees 9 1 6 Basic metrological terms c scccsccscsccsscssscsssecsscsessssssssscscsesssscsseessscssseeessoees 11 1 7 Structure OF The DOOKS isscessisesssissessosescsssoesesosssessseressncssssesstosssssisonssscasessessbnconssazeces 13 1 8 References nirera neet e eaa eea eas aa oe eo oa E oasa IEE 15 2 Review of thermal iMagers cccsseeeseeeeeeeeeeeeeeeeeeneeeeeeeeeeeees 18 2 1 Thermal cameras versus thermal scannerSs seessessessoesoesoossoesoesoossoesossoessssoosese 18 2 2 Applications of thermal cameras e ssseesessesoeseroesersesorsorsesorseroesorseroesoreororsorsoresoeee 19 PARE A E AET a o PE TA 20 2 4 Generations of thermal iMagerS seseeseesesersesorseeoesorseroesorsesorsoreerorsororeorseresoresoeee 20 Ze Technology trendS c csssscsecsascessserscssesesstsesasecessveossdnsssebadsosasivecsbesenssievassseseatosseenoss 27 PALA KAU S E A SE EEE EE E EE 28 3 Parameters of thermal imagers ss nnnssnnnnnnnsnnnnnnnnrennnnnn 29 3 1 Subjective image quality parameters e sessessesssoesossossscesosscecssossosscossossosesoesossoosses 31 SSL MIR N D EAA E A E E E E sbied restcns estes 31 Sel 2 MDT D os ieia a E EE als etn E A N eee 35 3 1 3 Evaluation of ran
219. seful when testing commercial measurement thermal imagers It can be estimated that absolute temperature range from 0 C to 100 C fulfills typical requirements for temperature range needed for testing surveillance thermal imagers and partially the requirements on blackbodies for testing commer cial thermal imagers Theoretically it is possible to use gray bodies during tests of thermal cameras and later to correct influence of difference between emissivity of the real blackbody and emissivity of an ideal blackbody on condition that the emissivity of the real blackbody is not lower than 0 9 due to problems with reflected radiation However a better option is to use real blackbodies of emissivity close to one when almost no correction is needed particularly during differential measurements Therefore emissivity not less than 0 97 or higher is considered nowadays as a standard re quirement Temperature range of some measurement thermal cameras goes over 1000 C and cannot be covered using thermoelectric blackbodies 97 Measurement of some characteristics is carried out for both positive and negat ive contrast Later the results are averaged Theoretically this approach enables full correction of systematic errors of measurement process mostly caused by offset phenomenon However practically this correction works properly only when blackbody does not change its temperature with time Practically this means that blackbodies of high te
220. sing parameter based on FPA sensor Table 3 8 Table 3 8 Specifications of exemplary IR FPAs and calculated resolution of thermal cameras No FPA Pixel Optics Number of DAS dimension focal detectors uv length Number of mm lines 1 320x256 HgCdTe 30x30 m 50mm 81920 pixels 0 6 mrad LWIR cooled 2 288x4 HgCdTe 28x25 m 50mm 288 lines 0 56 mrad LWIR cooled horizontal 0 5 mrad vertical 3 640x512 HgCdTe 15x15 m 50mm ___ 327680 pixels 0 3 mrad MWIR cooled 4 320x240 LWIR 45x45 m 50mm 76800 pixels 0 9 non cooled Resolution parameters based on FPA specifications can be misleading due to different reasons There are cases when a thermal camera built using IR FPA sensor of smaller pixel number can generate sharper image than a thermal camera built using IR FPA sensor of higher detector number This means that poor 640x480 pixel staring thermal camera can produce worse image that good 320x240 pixel thermal camera although such a situation occurs rather rarely Next when DAS is equal to x mrad it does not mean that we will be able to resolve 4 bar targets of bar width that equals x mrad There can be cases when 55 we will not be able to resolve even 4 bar targets of bar width three times bigger than the DAS Therefore the resolution parameters based on IR FPA specifications should be treated as indicators of imager theoretical ability to resolve small details More
221. sive anti vibration optical tables are not generally needed as a place where the test equipment is to be located The required accuracy of alignment of the test system is much lower than accuracy of alignment of some laser systems or holographic equipment In most cases the test equipment can be properly aligned on any large heavy and stable wooden stone metal table However special care should be taken during testing long range thermal imagers of very narrow field of view Vibration of the table can influence image quality of the image from the tested imager and can distort measurement results In this case it is recommen ded to use expensive optical anti vibration tables or to use modified typical tables with additional vibration damping parts 4 3 Manufacturing R amp D support equipment So far we have discussed in detail systems for testing thermal imagers understood generally as final ready products The tests using presented earlier test systems deliver a lot of information useful to verify final quality and to localize 109 weak points of manufactured thermal imagers that should be improved However other types of test equipment are needed for manufacturing thermal imagers and for research amp development projects Before a thermal imager is ready for final tests several important operations must be carried out 1 Correction of spatial noise 2 Calibration of measurement thermal imagers 3 Boresighting to a reference optical mec
222. stance thermometer readout Hart Scientific Wood L Laboratory Bench Analysis of Thermal Imaging Systems Opt Eng 15 G193 G197 1976 www ci systems com www ci systems com eo flir mets asp www eoi com www electro optical com datashts collimtr collimat htm www electro optical com eoi_page asp h Diffraction Limits www hgh fr www inframet com www inframet com testing infrared imaging systems htm www sbir com www optikos com www sbir com stc_collimators htm www sorl com productline oaps offaxisparabolas htm 115 5 Test procedures Methods of testing thermal imagers do not depend on test equipment used for measurements There are many similarities in procedures of measurement of para meters of thermal imagers using test equipment from different manufacturers but there are also some differences particularly resulting from the fact that nowadays most of these measurements are supported using specialized software 8 10 There fore in this chapter the simplified test procedures independent of software of test equipment will be presented These simplified test procedures refer directly to equipment manufactured only by one of manufacturers of test equipment Inframet www inframet com as only this equipment was at disposal of the author However the presented in this chapter test procedures should be valid also for test equipment from other manufacturers Over twenty parameters of thermal imagers can be measured to characterize
223. t all spatial frequencies In case of real imaging systems MTF always decreases to zero at some spatial frequency The shape of MTF function gives precise information about imager ability to produce sharp images In most imaging systems PTF is not significant and therefore it is usually assumed that OTF equals MTF Therefore MTF not OTF is typically used as a measure of quality of imaging systems Modern measurement methods of MTF of thermal imagers are based on rela tionships between MTF function and two other functions LSF and ESF MTF v Magnitude F LSF x 3 16 MTF v Magnitude F derivative from ESP x 3 17 where F is the Fourier transform operator LSF line spread function is one direc tional distribution of the flux in the image of an ideal line like target ESF edge spread function is one directional distribution of flux in the image of an ideal edge like target Measurement of MTF of thermal imagers is usually carried out on the basis of captured images of two types of targets narrow slit target or edge target Fig 3 27 When an image of one of these targets is captured and digitized then later MTF is calculated using formula 3 16 or formula 3 17 Practically measurement of MTF is not as simple as formulas 3 16 and 3 17 suggest due to necessity to use noise correction algorithm but these formulas present the principle of modern measurement of MTF of thermal cameras a Fig 3 27 Images captured d
224. t minimal movements of the imager can quite significantly im prove MRTD measurement results particularly in case of non cooled imager with high fixed pattern noise due to phase effects In case of staring imagers when the imager is slowly vibrating it is often easier to detect 4 bar pattern in comparison to situation when the imager is fully stabilized 5 1 8 Differential temperature steps Stanag 4349 Measurement of minimum resolvable thermal difference MRTD of thermal cameras that regulates measurement of MRTD of thermal imagers presents precise guidelines about steps of differential temperature that should be used during test procedure Table 5 4 These recommendations were perfect two decades ago when typical temperature resolution of thermal imagers was about 200 mK Regulation of differential temperature with 10 mK step is perfectly acceptable for such imagers However regulation of differential temperature with 10 mK step is totally not acceptable in case of modern thermal imagers particularly cooled staring imagers of temperature resolution often below 20 mK Next the recommendations about steps for temperature differences over 2 C are useless because this temperature range is not practically used during tests of modern sensitive thermal imagers Therefore the author updated the table with recommended differential steps to the form shown in Table 5 5 Table 5 4 Recommended steps of differential temperature during MRTD measurement
225. t the variable distance test system uses two 4 bar patterns bigger 10 mm bar smaller 4 mm bar International standards recommend to carry out MRTD measurements using tar gets with a single 4 bar pattern 1 4 The target can be rotated to have the bars in horizontal vertical or other position The aim of this is to help the observer to con centrate and make his task easier because he can expect to find only one 4 bar pat tern on a uniform background and there are no other objects that can distract atten tion of the observers However practical experience of the author shows that pla cing two 4 bar patterns one vertical 4 bar pattern and one horizontal 4 bar pattern do not reduce significantly accuracy of MRTD measurement but speeds up meas urement process as within one process both horizontal MRTD and vertical MRTD can be determined Therefore it is recommended to use double 4 bar targets in MRTD measurement process particularly when differences between horizontal MRTD and vertical MRTD are expected a Fig 5 2 Two 4 bar targets a single 4 bar pattern b double 4 bar pattern b 121 Table 5 2 Spatial frequencies of exemplary series of 4 bar targets in case of four vari able target test systems of different collimator focal length f bar width spatial fre spatial frequency spatial frequency spatial frequency mm quency 1 mrad 1 mrad 1 mra
226. tentially improved At the same time such tests should be carried out with high speed due to time restriction of typical production line Therefore MRTD measurement is done only for sample imagers in situation when measurement of MTF NETD FPN non uniformity FOV is carried out at production line In case of both surveillance thermal imagers and measurement thermal imagers three different test levels can be proposed basic typical expanded Table 3 10 The basic test level can be recommended for users of thermal imagers who are beginners to this technology and are looking for simple methods to verify quality of thermal imagers they purchased or to be purchased 69 Table 3 10 Recommended sets of parameters of thermal imagers Type of Test level Recommended set of parameters thermal imagers Basic MRTD Typical MRTD MTF SiTF NETD FPN non uniformity Surveillance A distortion FOV imagers Expanded MRTD or Auto MRTD MDTD MTF responsivity function NNETD FPN non uniformity distortion FOV 3D noise NPSD PVF Basic Minimum Error accuracy se aie Typical Minimum Error accuracy NGE SRF MRTD Expanded Minimum Error accuracy NGE SRF MRTD temperature stability The typical test level is recommended for more advanced users of thermal imagers who want to have detail information about thermal imagers due to different reasons One of such typical reasons is
227. tests of all short medium range thermal imagers and majority of long range thermal imagers Please note that such collimat ors are acceptable even in case of LW IR imagers of very narrow FOV built using very large optics 240 mm like Thermovision 2000 FLIR Inc Reflective collim ators of spatial resolution in the region 70 100 Ip mrad are recommended for test ing long range imagers of very narrow field of view designed using very large op tics and 640x480 or higher resolution FPA To summarize we can say that manufacturing of collimators for testing short medium range thermal imagers of wide medium field of view is relatively easy manufacturing good collimators for testing medium long range thermal im agers of narrow field of view much more difficult as the required spatial resolution Attention The warranty period of most of the tested collimators expired but the collimators are used usually many time over the warranty period 86 of the collimator is several times higher However a real challenge in field of re flective collimators is not to design a collimator to be used for testing typical sur veillance thermal imagers as listed in Table 4 4 but to design a collimator to be used for testing space CCD cameras The requirements on spatial resolution of collimat ors for testing such CCD cameras are at least five times higher than for collimators to be used even for testing best thermal imagers Design of collimators for testing
228. th development of polarization sensitive thermal imager to further improve the capab ilities of cooled technology 2 8 27 An increasingly noticeable trend appeared on the market to integrate thermal thermal imagers with other imaging and non imaging sensors Such integrated multi sensor surveillance systems thermal imager TV camera laser range finder have been used for quite a long time in airborne applications Nowadays however modern airborne imaging systems consist of more sensors high resolution four FOV thermal imager or two thermal imagers wide FOV color TV camera ultra narrow FOV color TV camera LLLTV camera laser range finder laser pointer laser designator laser illuminator 14 Next ground portable thermal imagers are more frequently integrated with additional modules like GPS laser range finder goniometer day light TV camera and laser pointer In some airborne naval or ground applications thermal imagers are integrated with classical radars or milli meter wave radars In all cases such integration significantly increases capabilities of thermal imagers 2 6 References 1 Baronti S et al Multispectral imaging system for the mapping of pigments in works of art by use of principal component analysis Applied Optics 37 1299 1309 1998 2 Beekman Daniel W Van Anda James B Polarization sensitive QWIP thermal imager Infrared Detectors and Focal Plane Arrays VI Conference SPIE Vol 4028 2000 3 Bel
229. the tested imager In both cases the tested imager generates distorted copies of the original targets images Next the images generated by the tested imagers are evaluated and important characteristics of the tested imagers are determined a b c d Fig 4 1 Images of standard targets used during testing thermal imagers a 4 bar target b pinhole target c edge target d slit target Technically we can formulate four basic requirements on a test system for test ing thermal imagers 1 Ability to simulate targets of different geometrical shape needed to measure different parameters of thermal imagers 2 Ability to regulate precisely angular size of simulated targets in order to simulate changes of distance at real conditions 73 3 Ability to regulate precisely temperature difference of the simulated target in comparison to background temperature needed in order to simulate variable contrast of thermal targets at real observation conditions 4 Ability to simulate targets located at distance bigger than minimal focus distance of the tested imager typical work conditions The requirements mentioned above can be fulfilled by several types test systems built using three different test principles 1 variable target test systems Fig 4 2 Fig 4 3 2 variable distance test systems Fig 4 4 Fig 4 5 Fig 4 8 3 variable target distance test system Fig 4 6 Fig 4 7 The variable target test systems Fig 4 2 Fig 4 3
230. ther on to simplify proced ure only MRTD test results for targets of vertical orientation are presented This simplified MRTD measurement algorithm is presented below 1 Write down basic test data imager type number and its field of view date of the test number and names of the observers ambient temperature As in the previous case we assume that ambient temperature equals 16 C correction coefficient 0 98 2 Calculate spatial frequencies of the 4 bar patterns at different distance imager test system or use data provided by the manufacturer of the test system like in the form shown in Table 5 3 3 Turn on the blackbody controller Set temperature difference equal to 2 C or more 4 Turn on the imager Put the imager at a short distance that can be considered as the minimal imager focus distance To determine this 137 10 11 12 13 14 15 16 17 18 19 20 21 distance please check imager specifications or simply reduce the distance until the image of the target becomes blurred Wait at least 15 30 minutes to enable both the imager and the test system to work in stable conditions Correct blackbody offset Set differential temperature to equal zero Regulate slowly this temperature around zero and determine the minimal positive temperature difference when the observer recognizes the target AT and the minimal negative temperature difference AT when the observer can recognize target Calculate offset a
231. thermal imagers night vision devices TV cameras laser range finders multi sensor surveillance systems non contact thermometry ond general metrology He is an author or co author of over 100 scientific papers and conference proceedings This books presents knowledge of the author on testing thermal imagers that was accumulated during over two decades of scientific work in the field of electrooptical technology interconnected with a series of practical projects The author hopes that this book will become a practical guide in a field of testing thermal imagers for wide community of people interested in this fascinating technology of thermal imaging ISBN 978 83 6 1486 8 1 7
232. ting thermal imagers is expensive Therefore take care that your funds are well spent on equipment truly useful in your work Manufactur ers can deliver or at least should be able different versions of the same test system of different measuring capabilities and at different price level Try to find a version of the test system that is optimal for your needs Third testing thermal imagers is difficult even having good test equipment par ticularly for newcomers to thermal imaging technology Test system shall generate measurement results but cannot automatically and properly interpret these results Such data must be interpreted by human users of test equipment who must do this difficult task We must remember that test results of a thermal imager often depend on imager settings environment conditions test equipment parameters etc There fore it is recommended to guarantee in the contract some support from equipment manufacturer covering also help in interpretation of test results Fourth buyers of equipment for testing thermal imagers often forget about problem of recalibration of test systems or rather about the problem of recalibration of some crucial blocks Costs of frequent recalibration of test equipment can be in long term almost equal to the original cost of purchase of this test equipment Therefore please check what are recommended recalibration intervals what is price of recalibration by the manufacturer and finally if it is possibl
233. tion of limit errors can be found in many books dealing with metrology Uncertainty theory proposes the uncertainty as such a measure of accuracy of meas urement results 16 Rules for evaluation of uncertainty in measurement are presented in the Guide to the expression of uncertainty in measurement 10 pub lished in 1993 by five main international metrological organizations the Interna 11 tional Organization for Standardization ISO the International Electrotechnical Commission IEC the International Organization of Legal Metrology OIML and the International Bureau of Weights and Measures BIPM Additional com ments can be find in Refs 6 and 11 The terms accuracy error systematic error random error uncertainty and limit error apparently seem to be easily understood intuitively However in practice these terms are often a source of confusion as it is possible to find radically different definitions in different literature sources Therefore let us define them clearly now to prevent any possible misunderstanding The International Vocabulary of Basic and General Terms in Metrology com monly abbreviated VIM published jointly by the mentioned above seven interna tional metrological organizations can be considered as the present day most import ant international standard 16 Definitions of five mentioned above terms accord ing to the VIM are presented below Accuracy of measurement VIM3 5 cl
234. tive nominal static ranges as the intersections of AT R and the converted MRTD functions Let us to practice using this algorithm by calculation of the detection ranges of the standard NATO target using a thermal camera of MRTD function shown in Fig 3 9 The calculation results are shown in Fig 3 10 and we can conclude that the de tection ranges are the following e 7 2 km at good atmospheric transmission e 2 8 km at limited atmospheric transmission Calculation of the recognition and identification ranges can be done in the same manner 2 5 2 15 a E 0 5 0 0 1 2 3 4 5 v 1 mrad Fig 3 9 Original MRTD measurement results Fig 3 10 Converted MRTD function for detection range and the functions A7 R at dif ferent atmospheric conditions rectangles MRTD values circle A7 R at good transmission triangles A7 R at bad transmission As we see algorithm provided by the STANAG 4347 enables to determine de tection recognition and identification ranges of the standard target using the tested 39 thermal cameras by doing only several non complicated mathematical operations The performance ranges of tested thermal camera in case of other targets of known parameters temperature difference and size of equivalent rectangle can be also de termined using the same algorithm The calculations can be carried out even using a simple calculator but can be s
235. tive off axis collimators of 150mm aperture and different F number F 5 triangles down F 6 6 squares F 10 triangles up To summarize if small size and mass is important then low F number collimators of short focal length should be preferred If high quality of the images projected by the collimator in its entire field of view is important then high F number collimators of long focal length are recommended However in most cases when the requirements on collimator spatial resolution are not high then both low F number collimators and high F number collimators can project images of sufficient quality and the test results get using both two types of collimators are the same C Spectral range transmittance IR collimator must projects thermal radiation emitted by the blackbody and targets located in its focal plane at least within the spectral range of the tested 7 Aberration blur was calculated as a angular diameter of a rectangle detector getting 71 energy of a ideal point source 89 thermal imager This means that spectral range of IR collimator must cover at least two spectral bands used in thermal imaging MWIR band 3 5 um and LWIR band 8 14 um preferable also the third band SWIR band 1 3 um Therefore we can conclude that for testing thermal imagers we need collimators of a spectral range that covers at least the region from 3 um to 15 um preferably from 1 um to 15 um If the collimator is to be used also for testi
236. to get temperature difference below zero to see negative 4 bar target However in case of test systems with a significant offset the negative target can be seen even for positive temperature differences close to zero see exemplary case in Table 5 7 Write the current temperature difference value as AT to Table 5 7 Decrease slightly the negative temperature difference until the 4 bar pattern disappears Exchange observers Increase slowly negative temperature difference until observer II starts to recognize 4 bar pattern Write the current temperature difference value as AT to Table 5 7 Decrease slightly the negative temperature difference until the 4 bar pattern disappears Exchange observers Increase slowly negative temperature difference until observer III starts to recognize 4 bar pattern Write the current temperature difference value as AT to Table 5 7 Exchange targets using the rotary wheel Target number 2 should be placed in the active position Repeat points 8 18 Repeat points 19 20 until the MRTD measurement was done for all 4 bar targets 130 26 Calculate MRTD values for each observer as an average from recorded positive temperature differences AT and negative temperature differences AT AT AT 2 Fill the AT columns for each observer in the test results table 5 6 27 Calculate raw measurement MRTDm as an average MRTD from each observer MRITD AAT AATot AT n 5 7 where MRTD is measureme
237. to zero Such a situation generates a constant bias in differential temperature readout There is a set of different factors that could create offset limited accuracy meas urement of target temperature using indirect method the temperature probe is not inserted into the target plate but into the rotary wheel wall difference of emissivity values of the target plate and blackbody radiator limited temperature uniformity on the surface of blackbody radiator etc If the MRTD measurement is done only for positive contrast patterns then offset can create significant measurement errors If the MRTD measurement is done for both positive and negative contrasts and the results are averaged then influence of the offset on MRTD results is corrected However this offset correction method is based on assumption that the offset does not change with time The time from the moment when MRTD measurement was done for positive contrast to the mo 124 ment when MRTD measurement was done for negative one can be estimated as 2 4 minutes Offset can change quite significantly during that time if tests are car ried out at room of low stability of ambient temperature and the test system is not protected well against ambient temperature variability As we see the effectiveness of the offset correction method depends directly on temporal stability of the temperature difference between the target and the black body indirectly on quality of a test system and the en
238. ts Most manufac turers of equipment for testing thermal imagers provide also computer programs that help human observer to carry out MRTD measurement In detail the tasks of these support programs are following l Nn Recording measurement conditions imager name or other identification data number of observers test date test place ambient temperature col limator transmittance collimator focal length test orientation Calculation of spatial frequency of 4 bar targets in lp mrad on the basis of known bar width in mm unit Correction of influence of measurement conditions on test results correc tion of influence of ambient temperature and of collimator transmittance User friendly control from PC of blackbody temperature and position of rotary wheel Presentation of MRTD measurement results in form of tables and graphs Recording test results 140 MRTD test conditions Imager name number and FOV no name Number of observers 4 Test date 2007 03 27 Test place Jundefinded m Define targets fxi Test system type and number No Bar width mm Jundefined 4 Ambient temperature 20 2 38 2 Colimator transmittance 1 004 E T Colimator focal length 0504 1 059 m Target position Horizontal Vertical t iP Define targets elias Cooled Mw 7 J Symmetric MATD fed Export to CSV is Clear table Fig 5 10 Windows of TCB Control program used for suppo
239. ts of different spatial frequency at the same temperature difference a b Fig 3 3 Image of 4 bar target at two different temperature differences 32 _ 8 10 12 2 l Il Il l mi spatial frequency lp mrad Fig 3 4 MRTD of exemplary thermal camera of three different field of view Classical MRTD is a subjective parameter that takes the observer into account This subjectivity and time consuming measurements cannot be accepted in high volume production environment where so called AutoMRTD is preferred AutoMRTD is a test methodology that proposes a quick objective method to measure MRTD of a thermal camera using an algorithm presented below 1 Measure MRTD NETD noise equivalent temperature difference and MTF modulation transfer function of a large sample of thermal cam eras 2 Calculate average MRTD NETD MTF for the tested sample cameras 3 Next calculate the coefficient function K v as _ MRTD v MTF v ey NETD 3 1 4 Calculate new objective MRTD of any new tested thermal camera using the following formula K v NETD MRTD auol TF In order to determine objective MRTDwto it is required to measure only two objective parameters NETD noise equivalent temperature difference and MTF modulation transfer function Measurement of both MTF and NETD is fast and semi automatic Therefore the measurement time needed to determine objective MRTDauto is much shorter than the time needed to me
240. tted radi ation we can measure parameters that describe precisely performance of different infrared imaging systems thermal cameras NVD and different types of TV camer as In spite of the same general testing concept there are significant differences between the apparatus for testing far infrared imaging systems thermal imaging systems and the apparatus for testing visible near infrared systems NVD TV cam eras Photometric methods developed visible range are used also for near infrared range when radiometric methods based on temperature control are needed for far infrared range List of differences in the test methods for testing the far infrared systems and the test methods for testing visible near infrared systems is presented in Table 1 1 Table 1 1 Basic differences between apparatus for testing far infrared systems and ap paratus for testing visible and near infrared systems EY rca cameras bod measurement temperature quantities tics units of radiation standard temperature units or radiometric units targets transparent patterns non transparent pat holes in metal sheets terns on transparent glass sheets 1 5 Terminology In spite of a relatively long tradition of IR systems still there are no internation ally accepted terminology standards in most areas of this technology At present only terminology related to quantities of infrared radiation and detectors of this ra diation has been relatively well st
241. ture differences measured for a single contrast for each of the remaining spatial frequencies in the test run Positive con trast measurements for these frequencies are recommended 14 Note 4 It is proposed that the black body Most blackbodies used for testing temperature be varied in steps as follows thermal imagers enable temperat ure control with 0 001 C steps much better than required lt 0 5 C 0 5 C to 1 C 6 2 It is recommended that the MRTD be Most rotary wheels are designed measured with horizontal and vertical ori in a way that enables the 4 bar entation of the bars Any other orientation targets to be fixed by the user in is optional vertical orientation or horizontal orientation 6 3 The criterion to be used for resolu This is an requirement to be ful tion is that it should be possible to see filled by users of the test equip four bars not just some modulation on ment not requirement on the 160 the display although it is not necessary that all of each of the four bars be visible at the same time 6 4 Measurements shall be taken at a minimum of four 4 spatial frequencies distributed approximately uniformly over the useful range of the imager A minim um of three 3 measurements shall be taken for each pair of contrasts at each spatial frequency in each orientation by each observer Measurements shall be taken until stable results are observed A single observer may be us
242. ty in measurement guidelines on evaluation of limit error in numerous metrology handbooks To summarize the uncertainty is nowadays the measure of measurement accur acy recommended by international metrological organizations and there are quite clear rules how the uncertainty of measurements should be calculated However the problems in consistent use of these recommendations are that these recommend ations are often ignored by international community working in the field of electro optical metrology due to different reasons 1 7 Structure of the book As it was stated in Section 1 2 electro optical imaging systems can be divided into three basic types night vision devices TV cameras and thermal imagers This book is devoted to the problems of testing and evaluation of thermal imagers There are many myths about thermal imagers The conflicting opinions about performance of these modern and fascinating imaging systems are given It is also known that quality of thermal imagers offered on the market vary significantly The only way to be sure about quality of the interesting imagers is to test them and evaluate test results Nowadays testing thermal imagers is very important for many people involved in thermal imaging technology due to several reasons Firstly thermal imagers are still quite expensive A the same time the number of thermal imagers used in surveillance applications by military police border guards rescue teams et
243. umber and big off axis distance is needed Therefore the reflective targets are met rather rarely when testing thermal imagers mostly in case of collim ators of modified design with improved baffling The tested imager can see also itself reflected on the surface of the reflective target 102 BLACKBODY 1 PRIMARY MIRROR FOLD MIRROR TARGET WHEEL BLACKBODY 2 E Fig 4 20 Block diagram of test system optimized for use of reflective targets and two blackbodies 3 Targets of different shape are manufactured to enable measurement of different parameters of thermal imagers We can distinguish at least fourteen shapes of IR targets 1 Four bar targets 8 Double 4 bar targets 2 Pinhole targets 9 Multiple 4 bar targets 3 Square targets 10 Multiple pinhole targets 4 Slit targets 11 Abingdon cross targets 5 Interlace targets 12 Distortion targets 6 Edge targets 13 Grey scale target 7 Alignment targets 14 Silhouette targets Drawings of these targets are shown in Fig 4 21 and their description in Table 4 7 1 2 3 4 6 7 5 12 Fig 4 21 Drawings of different types of IR targets 103 Table 4 7 Application area of different types of IR targets No Target Description Application Comments type 1 4 bar single 4 bar pattern MRTD a set of targets with various spatial 7 1 bar frequencies is needed
244. urements are carried out first for positive contrast targets and later for negative contrast targets Blackbody accuracy is important during calibration process of measurement commercial thermal cameras when the blackbody is used as a reference standard Absolute accuracy of modern commercial thermal cameras is typically 2 or 2 C whichever higher in camera temperature range Accuracy of blackbodies should be at least ten times better Therefore it seems that temperature errors of blackbodies at a level of 40 2 or 0 2 C should be considered as acceptable If the manufacturer can deliver blackbodies of better accuracy it should be welcome 98 Table 4 6 Summary requirements on blackbodies Parameter Acceptable Recommended Temperature 10 mK for testing non 1 mK typical situation resolution cooled imagers 0 1 mK extreme cases 1 mK for testing cooled III gen imagers Absolute 15 C 35 C at 25 C 0 C 100 C _ typical Temperature range ambient temperature situation 15 C 100 C extreme situation Differential 10 C 10 C at 25 C 25 C 75 C typical Temperature range situation 40 C 75 C extreme situation Emissivity 0 96 20 97 Settling time 10 C 120s at 10 mK level 60s at 10 mK level step seconds Uniformity 20 mK for 5 C 10 mK for 5 C temperature temperature range range 150 mK for 30 C 100 mK for 30 C tempe
245. uring MTF measurement a image of a narrow slit target b image of an edge target b gt If case of a linear PTF only simple lateral displacement of the image is observed Non linear PTF can adversely affect image quality An extreme case is a phase shift of 180 produces a reversal of image contrast 59 MTF function is an excellent criterion of image quality of thermal cameras However interpretation of a curve is more complicated than interpretation of simple numerical parameters Next graphical presentation of MTF function was difficult several decades ago when computers were rarely used Therefore several numeric al parameters based on MTF function were proposed in the past to characterize thermal cameras Nowadays these numerical parameters are rarely used but still it is useful to know these five numerical parameters related to MTF function 1 Equivalent frequency equivalent line number or equivalent bandwidth N 2 Half frequency HF 3 Effective resolution ER 4 Effective instantaneous field of view EIFOV 5 Limiting resolution LR First the equivalent frequency Neis defined as 00 N MTF dv 0 3 18 where N is the equivalent frequency called also equivalent line number or equivalent bandwidth XN is presented using spatial frequency units The equivalent frequency N concept is based on Shade criterion 24 who stated that perceived image quality can be described using the formula 3 18 Secon
246. valuation of Thermal Imagers For Non Destructive Thermal Test ing Applications Infrared Physics and Technology 42 2 101 105 2001 6 Chrzanowski K Szulim M A measure of influence of detector noise on temperature measure ment accuracy with IR systems Applied Optics 37 5051 5057 1998 7 Chrzanowski J Fischer W Wrona Testing of Thermal Imagers For Non Destructive Thermal Testing Applications ASTM Journal of Testing and Evaluation 28 395 402 2000 8 D Agostiono Webb C 3 D Analysis Framework and measurement methodology for infrared systems noise Infrared Imaging Systems Design Analysis Modeling and Testing SPIE Vol 1448 110 121 1991 9 Driggers Ronald G Van A Hodgkin Vollmerhausen Richard H O Shea Patrick Minimum resolvable temperature difference measurements on undersampled imagers Proc SPIE Vol 5076 Infrared Imaging Systems Design Analysis Modeling and Testing XIV 2003 10 Guide to the expression of uncertainty in measurement International Organisation for Standarisation International Electrotechnical Commission International Organisation of Legal Metrology International Bureau of Weights and Measures TAG 4 WG 3 1993 11 International Vocabulary of Basic and General Terms in Metrology International Organisa tion for Standarisation 1993 12 Holst G C Infrared Imaging System Testing Vol 4 Chapt 4 in The Infrared amp Electro Optic al Systems Handbook Michael C Dudzik ed SPIE 1993
247. vironmental conditions in the test room Therefore it is highly recommended to use for MRTD measurement of modern cooled staring thermal imagers the highest requirements only test sys tems that of highly reduced offset variations If not the measurement results at low spatial frequency range shall not be repeatable If ofset stability is confirmed then we have an opportunity to speed up measure ment process We can first correct the offset and next carry out MRTD measure ments only for positive contrast In case of most test systems offset can be easily measured and later eliminated using blackbody controller It can be measured and later calculated using this for mula offset AT AT_ 2 l 5 5 AT is the minimal positive temperature difference when the observer recognizes the target and AT is the minimal negative temperature difference when the observer can recognize the target When offset was calculated using the formula presented above the observer should insert offset value into electronic memory of the controller Then the bias in indications of differential temperature is eliminated and we can consider the test systems as offset free at least for some limited time 5 1 6 Imager settings Observers are allowed to regulate all imager settings until they consider that they get the best image during MRTD measurement Here we present several guidelines a set gain contrast to maximum during measurements when low fre
248. vity to ambient temperature objects and its better transmission through mist and smoke However the 3 5 mm band may be more appropriate for the hotter object or if sensitivity is less important than con trast Also additional differences occur e g the advantage of MWIR band is smaller diameter of the optics required to obtain a certain resolution and that some detectors may be operated at higher temperatures thermoelectric cooling than it is usual in the LWIR band where cryogenic cooling is required about 77 K Therefore there is no definite always valid answer which type of thermal imagers MW thermal im agers or LW thermal imagers should be preferred Both types of thermal imagers have certain advantages and disadvantages 2 4 Generations of thermal imagers Thermal cameras are generally divided into three generations Scanning cameras built using discrete detectors simple non multiplexing photoconductive linear ar 20 rays typically PbSe InSb or HgCdTe of elements number not higher than about one hundred or the SPRITE detectors are the first generation thermal cameras They usually operate in 8 12 um spectral range use the optics of F 2 F 4 number and are characterized by temperature resolution NETD about 0 2 K Small quantit ies of first generation thermal cameras were introduced as military equipment in the 1970s more in the 1980s Thousands of these systems are still in military ser vices spare part will be available for
249. wo bars can be wider than the others but still the observer must be able to recognize separately four bars VIRTEST computer program freely available from www inframet com website can be used for training 117 5 1 3 Test environment Observation conditions like ambient illumination ambient temperature can significantly influence measurement results It is generally considered that ambient illumination conditions during the MRTD tests should be similar to normal working condition of the tested thermal imager Next the recommended ambient temperature by the test standrds is 20 2 C If tests are carried out at different temperatures then the test results should be corrected Influence of two factors on measurement results should be taken into account and corrected ambient temperature and transmittance of the optical channel between imager and the test system Ambient temperature depending on exact value can both decrease or increase MRTD measurement results Limited transmittance always increases measurement results Influence of ambient temperature on measurement results can be corrected by multiplying the measurement results by a certain correction coefficient cor In case of non cooled thermal imagers it can be assumed that imager noise does not depend on ambient temperature the coefficient can be determined using the following formula dM A T V da ysl d aT sys h IM A T OMG Tracey wo sh dh cor T
250. ximal sharpness of image gen erated by the imager The sharpness can be evaluated subjectively by hu man eye but use of objective evaluation tool is recommended Some manu factures of test equipment offer software modules called Live MTF or 145 Continuous MTF that enable online measurement of sharpness of edge image generated by the tested imager 8 10 User is only requested to mark a part of the edge to be analyzed Due to significant noise in analyzed images accuracy of Live MTF measurement is high and true measurement of MTF must be made later using some noise reduction methods i a a b Fig 5 12 a Image of the edge target with marked area to be analyzed b MTF graph 5 Determination of relationship between image dimension in pixels and true dimension in angular units mrad Knowledge about the edge target di mensions and the collimator focal plane is used to find this relationship This operation is often called Distance Calibration The operation is carried automatically by software or the user is expected to mark some points on the image generated by the tested imager 6 Capturing a video sequence of images of the edge target A series of images is captured in order to reduce temporal noise by simple averaging method Some test computer programs use also more advanced image processing techniques that can reduce additionally high frequency spatial noise fixed pattern noise 7 Analyzing the image o
251. y AA 0 ba s where Tsa is the ambient temperature for which the measurements were made Tac is the standard background temperature M T A is the spectral exitance at the temperature T and the wavelength sys A is the imager relative spectral sensitivity and 44 is the imager spectral band In case of cooled thermal imagers it is assumed that BLIP detectors of ambient temperature depending on noise are used the coefficient cor can be determined using this formula MO T dh Ad d ba IM A T ZON dl Ad d ba s cor T 5 2 If we assume that the reference ambient temperature equals 20 C then for typical thermal imagers of the 3 5 um and the 8 12 um spectral bands the correction coef ficient cor can be presented as shown in Fig 5 1 118 Coefficient 0 4 0 2 0 5 10 15 20 25 30 35 40 45 temperature C Fig 5 1 The correction coefficient cor for typical thermal imagers of the spectral bands 3 5 um and the 8 12 um squares non cooled 3 5 um rhombus non cooled 8 12 um triangle up cooled 8 12 um triangle down cooled 3 5 um Using the values of the coefficient cor shown in Fig 1 1 and Table 1 1 it is theoretically possible to correct MRTD measurement results when measurements were carried out at a range of ambient temperatures from 0 C to 40 C However the real thermal imagers fulfill relatively well assumptions taken to deriv
252. y low Table 4 3 suggests that resolution of even big collimators aperture about 250 mm during tests of LW thermal imagers is below 5 lp mrad due to diffraction limit of the collimator This means that using typical off axis reflective collimators for projecting images of targets of frequencies over 5 lp mrad we should always get blurred images of these targets generated by the tested LW thermal imagers Such situation should occur even if the tested im agers is perfect because the collimator is the limiting factor It is not true as the au thor of this book tested some long wavelength thermal imagers that generated sharp images of targets of frequency over 10 Ip mrad clearly over the suggested diffraction limit of the collimators The situation described above is possible because the manufacturer 28 used to calculate the values of limited target frequency Vmax using the most pessimistic for mula met in literature The limited target frequency Vmax is calculated generally us ing this formula D cm 1 mrad x uum v max 4 1 where D is the collimator aperture is the wavelength and x is the coefficient The manufacturer 28 used the coefficient x equal to 2x2 44 4 88 in a situation when it is possible to find literature sources where the x coefficient equals one Therefore the values of diffraction limited target frequencies presented in Table 4 3 are so low The practical consequence of using formula 4 1 or the web calcu

TESTING THERMAL IMAGERS - Practical Guide

Contents

Download Pdf Manuals

Related Search

Related Contents