Home

Software Development Kit User's Guide

1. ees 3 1 MuTech Product Support 3 1 Hierarchical zd RUE B 8 Huffman code tables 1 2 Huffman coding oo oooo o 1 2 1 4 A 1 B 5 Huffman decoding 0 02 ee ee eee A l Huffman oble ccc eee A 1 B 5 interleave 3 ghey er eee So IN a eds B 6 Inverse DCT 0 eee 1 3 IV 4XX SDK o o oooo o 2 1 2 3 3 30 3 32 IV 4XX Video Frame 3 30 3 31 3 33 3 34 Joint Photographic Experts Group 1 1 A 1 B 1 JPEG 5 vb pue PREPARED TESTES Y 1 1 B 2 Library DOS EE 2 1 TOSSI OS Strat K oen See BORO SES 1 1 3 16 3 35 LOSSY EE 3 16 Microsoft Visual CC 2 1 2 3 neighborhood prediction 1 3 neighboring samples C 1 non interleave llle B 6 original image 0 cee eee eee eee 1 4 Version 2 0 P pixel Tomat 3 24 A 1 B 1 C 1 prediction d Terence 0 02 00 ee eee 1 4 prediction error 2 2 2 eiaa ee eee 1 3 predictor 3 1 5 3 5 3 9 3 15 3 23 3 36 A 2 B 3 C 1 one dmenstonal eene C 2 Q Quality Factor 1 1 1 3 3 9 3 15 3 16 3 23 3 24 3 36 Quantization llle 1 2 A2 Quantization Goble aa nannaa 1 2 B 4 R Region Of Interest 3 14 3 19 3 22 3 28 RGB ceded tee bk ak 1 2 3 2 3 7 3 16 3 39 3 42 ROL La Deu Delos debi NS cci ier SACH 3 14 3 28 SDK VETSION aec e Ee 3 2 SOUICE sue ei uS een NS 3 2 SubSampling 00 0002 A 2 T time stamp
2. SIDE EFFECT SEE ALSO EXAM PLE Version 2 0 PARA_ERROR One of the parameters is invalid This is functionally identical to JPEG File2File Compress The only difference is that this function stores the compressed code stream into a memory buffer instead of a disk file The codes generated by these two functions are identical Please refer to JPEG File2File Compress for more comments None JPEG File2File Compress JPEG Memory2File Decompress MTInquireJPGInMemoryVals CompressParameter cPara BYTE HUGE buffer UWORD dx dy bp ULONG length fill out the input parameter structure cPara CompressMethod JPEG LOSSY cPara QulityFactor 128 cPara CompressedDataFormat JPEG AS SOURCE cPara Comments NULL cPara CommentLength 0 assume the compression ratio is greater than 1 1 the following buffer length is considered most conservative that assume nothing is compressed MTInquireBMPVals test bmp amp dx amp dy amp bp bp bp 8 bytes per pixel length ULONG dx ULONG dy ULONG bp allocate buffer buffer BYTE HUGE compress if JPEG_File2Memory_Compress test bmp cPara buffer amp length TMALLOC length OK error occurs Now the length contains the actual size of the code stream in bytes decompress if JPEG Memory2File Decompress buffer length DATA AS COMP
3. DATA_16bit_YCbCr422 bpp 2 else dFrame PixelFormat DATA 8bit BW bpp 1 allocate memory dFrame buffer BYTE HUGE MTMALLOC ULONG dFrame dx ULONG dFrame dy ULONG bpp dFrame UpDownFlip 0 fill out the region of interest structure here the ROI is the same as the Data Frame roi sx 0 roi sy 0 roi dx dFrame dx roi dy dFrame dy decompress if JPEG_File2Memory_Decompress test jpg dFrame roi NULL NULL OK error occurs the decompressed data can be processed her fill out the input parameter structure cPara CompressMethod JPEG LOSSY cPara QulityFactor 128 cPara CompressedDataFormat JPEG AS SOURC cPara Comments NULL cPara cLength 0 Gl compress JPEG Software Development Kit User Guide Version 2 0 3 18 Function References if JP MTFRE I D EG Memory2File Compress dFrame roi cPara newtest jpg OK error occurs dFrame buffer 3 1 8 JPEG File2Memory Decompress SYNOPSIS DESC RIPTIO N PARAM ETERS Version 2 0 inclu de mtjpeg h int MTPROCALL JPEG_File2Memory_Decompress BYTE FAR InFile DataFrame dataFrame ROI region BYTE FAR Comments UWORD FAR cLength Decompresses an image file in JPG format into a region of interest within a Data Frame InFile A pointer to a string that
4. JPEG AS SOURC cPara Comments NULL cPara CommentLength 0 Gl compress if JPEG File2File Compress test bmp cPara test jpg OK error occurs decompress if JPEG File2File Decompress test jpg DATA AS COMPRESSED test tga NULL NULL OK error occurs 3 1 4 JPEG File2File Decompress SYNOPSIS DESC RIPTIO N PARAM ETERS Version 2 0 include mtjpeg h int MTPROCALL JPEG File2File Decompress char FAR InFile BYTE PixelFormat char FAR OutFile BYTE FAR Comments UWORD FAR cLength Decompresses an image file in JPG format into an image file of BMP TGA or TIF format InFile A pointer to a string that specifies the input file name which may contain a path and must have an extension of JPG PixelFormat This parameter must be filled with one of the pre defined macros DATA AS COMPRESSED Used for most cases see comments for details DATA l6bit YCbCr422 JPEG Software Development Kit User Guide Function References 3 7 This is an exception that may be used only when the compression was lossy and the OutFile has an extension of TIF See comments for details OutFile A pointer to a string that specifies the output file name which may contain a path and must have an extension of BMP TGA or TIF lt Comments Pointer to a byte buffer that will be used to hold the JPEG comment information if one e
5. UWORD sy UWORD dx UWORD dy UWORD flag BYTE CompressMethod BYTE QualityFactor BYTE FAR Comments UWORD cLength Compresses an ROI in a pre defined IV 4X X Video Frame and saves into a JPG file name specifies the file name The file name can contain a path and must have an extension of JPG SX Sy specifies the start point offset of the ROI horizontally sx and vertically sy in the pre defined IV 4XX Video Frame source dx dy specifies the size of the ROI When sx dx is greater than the width of pre defined IV 4X X Video Frame a clipping will occur The dx or dx if clipping happened is the size of the compressed image The dy is working the same way on vertical direction flag specifies how the image should be saved it could be one of the pre defined macros IV4 Save As Is The image should be saved as is If theimage is in B W mode it will be compressed in Black White format Color image will be in YCbCr422 format IV4 Save As BW The image should be compressed in Black White format no matter what the original image format is IV4 Save As YUV The image should be compressed in YCbCr422 format no matter what the original image format is If necessary the color space conversion is applied automatically This is detailed in the following table JPEG Software Development Kit User Guide Function References 3 35 flag IV AXX On Board Compressed F
6. specifies the start point offset of the sub area relative to the upper left corner of the image horizontally sx and vertically sy in the named JPG file source dsx dsy specifies the start point offset of the ROI horizontally dsx and vertically dsy in the pre defined IV 4XX Video Frame destination dx dy specifies the size of the sub area If no clipping happened the sub area sizes for the source and the destination should be the same When sx dx is greater than the width of the compressed image a clipping will occur in the source side which produces a new dx When dsx dx is greater than the width of the IV 4XX Video Frame a clipping will occur in the destination side The sy dy and dsy are working the same way on the vertical direction flag specifies how the image should be loaded It could be one of pre defined macros IV4_Load_As_Is The image should be loaded as is If the image is in 8 bit format it will be loaded in Y only Black White format if the image is in color format it will be loaded in 16 bit YCbCr422 color image IV4 Load As BW The image will be loaded in Y only Black White format no matter what the compressed image is If necessary the color space conversion is applied automatically This is detailed in the following table flag Compressed Format IV 4XX On Board Format IV4_Load_As_Is Black White 8 bit B W Cb Cr 0x80 IV4_Load_As_Is Color Color 16
7. Format is selected in the Data Frame If the Pixel Format in the Data Frame is DATA_16bit_YCbCr422 or DATA_16bit_YONLY the image data will be converted to DATA_8bit_BW before compression JPEG_16bit_YCbCr422 Specifies that the Compressed Data Format will be 16 bit per pixel YCbCr422 format This parameter is valid only when the input data format is DATA 16bit YCbCr422 JPEG AS SOURCE The Compressed Data Format will be determined by input data i e Pixel Format in the Data Frame When the Pixel Format is DATA 8bit BW the Compressed Data Format will be JPEG 8bit BW When the Pixel Format is DATA 16bit YCbCr422 or DATA 16bit YONLY the Compressed Data Format will be JPEG 16bit YCbCr422 B Comments Pointer to a byte buffer that contains application related information such as comment time stamp etc The contents in the buffer will be copied to the compressed jpegBuffer using a JPEG Comment Marker The user can retrieve this information by calling either the MTInquireJPGInMemory Vals or one of the decompression functions that decompresses from a JPG memory buffer E CommentLength Specifies the length of the information in bytes If the contents is longer than CommentLength the extra part will be dropped The user must ensure that the CommentLength is not longer than the size of the Comments buffer amp jpegBuffer A pointer to a buffer of bytes which will be used to store the compressed code stream in JPG format
8. JPEG SDK provides a 32 bit Dynamic Link Library DLL MTJPEG32 DLL and an import library MTJPEG32 LIB which are based on Microsoft Visual C C 2 x or later and chich work under both Windows 9X and Windows NT Since MTJPEG32 DLL is a standard 32 bit Windows DLL it should work with other compilers as well Once installed the following files can be found in the sub directories X MTJPEGNINCV MTJPEG H PORTABLE H STD_DEF H X MTUPEG LIB MTJPEG32 LIB MTJPEG32 DLL X MTJPEGNDOCNREADME W32 X MTJPEG SAMPLES JPEG WIN32 CF2FCMP C CF2FCMP MAK CF2FCMP DSW CF2FCMP DSP CF2FDEC C CF2FDEC MAK CF2FDEC DSW CF2FDEC DSP X MMTJPEGNSAMPLESNJPEGNWIN32N C RY C RY H RY RC RY DEF RY MAK RY DSW RY DSP OOOOOcOoO HEE Ew 000 000 03 The following files are also included in the MuTech JPEG SDK but are ex pected to be used with the IV AX X SDK for Windows 9X or for Windows NT JPEG Software Development kit User s Guide Version 2 0 2 4 Using MuTech JPEG SDK X MTJPEGNSAMPLESNIVAJPEGNWIN95N CJPEG C CJPEG H CJPEG RC CJPEG DEF CJPEG MAK CJPEG DSW CJPEG DSP E E Dj aa Q Q Hew Q Q Q IV4JPEG WINNT Ej X MTJPEGNSAMPLESN J C H z C CE E mb pn ops aa Q RE DEF MAK DSW DSP Z q m t Q z C oos Ej E DD E Q Q Q Please r
9. VALUE COMMENTS SEE ALSO Version 2 0 include mtjpeg h void MTPROCALL MTYCbCr422ToRGB32Bit void HUGE input void HUGE output ULONG length UWORD RGBOrder UWORD SubSample Converts a stream of YCbCr data to a stream of RGB data input Pointer to the input buffer which contains YCbCr data to be converted The length of the input buffer must be at least 2 times the number of pixels length 2 The data format of the input buffer is YCbCr422 YUYV where each unit corresponds to each pair of RGB pixels output Pointer to the output buffer which is used to store the results of the conversion The length of the output buffer must be at least 4 times the number of pixels length 4 The data sequence of the output buffer is defined by the RGBOrder parmeter length Specifies the number of pixels to be converted The length must be even number RGBOrder Specifies the order of the RGB data strored in the output buffer When RGBOrder is 1 the output data format is RGBXRGBX when RGBOrder is O the output data format is BGRXBGRX where X is the byte discarded in the conversion SubSample Reserved and must be set to 0 None Also see comments for JPEG_File2Memory_Decompress MTRGB32BitToYCbCr422 JPEG File2Memory Decompress JPEG Software Development Kit User Guide Function References 3 43 3 3 5 MTInquireJPG Vals SYNOPSIS DESC RIPTIO N PARAMETERS include mtjpe
10. allocate a buffer which is large enough to hold the maximum anticipated size of the comment field then calls this function to get the information needed None JPEG Memory2Memory Compress MTInquireJPGInMemoryVals JPEG File2Memory Decompress See sample for JPEG Memory2Memory Compress 3 2 Mulech Products Support Group 3 2 1 IV AXX Board Support Group The functions in this group must be used with IV 4XX SDK The function prototypes for this group can be found in IV4 H provided by IV 4XX SDK Both MuTech JPEG DLLs and corresponding IV 4XX SDK DLLs must be present for these functions to be called successfully For functions described in this sub section please also refer to the document IV 4XX Software Development Guide For functions in this group unless specified otherwise all coordinates and dimensions are in unites of pilxes and lines 3 2 1 1 IVAJPEGLoad SYNOPSIS DESC RIPTIO N Version 2 0 include iv4 h int MTPROCALL IV4JPEGLoad char FAR name UWORD sx UWORD sy UWORD dsx UWORD dsy UWORD dx UWORD dy UWORD flag BYTE FAR Comments UWORD FAR cLength Decompresses a sub area of the image in a JPG file and loads into an ROI in a pre defined IV AX X Video Frame JPEG Software Development Kit User Guide Function References 3 31 PARAMETERS name specifies the file name The file name can contain a path and must have an extension of JPG Sx SY
11. buffer pointed by Comments and return the actual length of the comment field in the variable pointed to RETURN VALUE OK The function completed successfully PARA ERROR One of the parameter is invalid COMMENTS In most cases the DATA AS COMPRESSED is used for PixelFormat The Decompressed Data Format will be determined by how the data was compressed in the InFile If the Compressed Data Format is 8 bit Black White the Decompressed Data Format will be 8 bit Black White If the Compressed Data Format is 24 bit RGB the Decompressed Data Format will JPEG Software Development Kit User Guide Version 2 0 3 8 Function References SIDE EFFECT SEE ALSO EXAM PLE be 24 bit RGB for lossless Or for the Compressed Data Format of YCbCr422 the Decompressed Data Format will be converted to RGB for lossy except if the OutFile is TIF and if the compression was lossy then the output format can be the YCbCr422 format which must be specified by using the pre defined macro DATA_16bit_YCbCr In General there two ways that the application can retrieve the comment field from a compressed code stream The first is to call MTInquireJPG Vals function to find out the information about the comment field then allocate the buffer of the right size and retrieve the comment field during the decompression function call The second way is to blindly allocate a buffer which is large enough to hold the maximum anticipated size of the comment fie
12. bufferLength A ULONG variable that specifies the length of the buffer The requirement of the buffer length depends on the compression ratio which is in turn depending on the Quality Factor parameter It is the caller s responsibility to allocate enough space to accommodating the code stream A rule of thumb is to allocate a buffer the same size of the original image RETURN VALUE OK The function completes successfully Version 2 0 JPEG Software Development Kit User Guide Function References 3 25 PARA_ERROR One of the parameter is invalid COMMENTS This function is functionally identical to JPEG Memory2File Compress 1 The only difference is that this function stores the compressed data into a system memory buffer instead of a disk file The data generated by these two functions are identical Please refer to JPEG_Memory2File_Compress for comments SIDE EFFECT None SEE ALSO JPEG Memory2Memory Decompress MTInquireJPGInMemoryVals JPEG Memory2File Compress EXAMPLE cCompressParameter cPara DataFrame dFrame BYTE monoInfo bp BYTE HUGE jpegBuffer ULONG length dFrame buffer BYTE HUGE MTMALLOC ULONG dFrame dx ULONG dFrame dy ULONG monoInfo dFrame UpDownFlip 0 fill the data frame with 8 bit black white data from other source dFrame PixelFormat DATA 8bit BW fill out the region of interest structure roi sx 0 roi sy 0 roi dx dFra
13. one exists in the compressed code stream A NULL pointer could be used if the user has no interest of the comment content at the time of calling If a buffer pointer is specified its length must be specified by cLength see below for detailed description It is the caller s responsibility to allocate enough space for the buffer otherwise the extra part of the Comments field will be discarded The user can also retrieve the comment information by calling MTInquireJPG Vals function JPEG Software Development Kit User Guide Version 2 0 3 38 Function References cLength Pointer to an UWORD variable that can be used in a number of ways described below E points to NULL the user is not interested in any information in the comment field The function will not look for Comment Marker and totally ignore the Comments buffer E points to a variable with the value 0 The user is interested in only the length of the comment field If there is no Comment Marker found the return value will be 0 unchanged Otherwise this function will return the actual length of the comment field in the variable pointed to The Comments buffer will be ignored E points to a variable with non zero value The user is interested in both the length and the content of the comment field In this case the Comments buffer must be allocated and supplied and the value of the variable must be equal to the size of the Comments buffer If there is no Comment Marke
14. rithm This method uses a 3 pixel predictor with a lossless encoding of the prediction error B 3 Basic DCT Block Diagram Common to all encoding modes of the DCT based algorithm are the compo nents shown in Figure B 1 which illustrates the key processing steps applied to each 8x8 block of samples from a single component of a source image The encoder must extend the right and bottom edges of the image component so that it will consist of an integral number of blocks The actual dimensions of the image are encoded in the header and the decoder in turn discards the ex tensions JPEG Software Development kit User s Guide Version 2 0 B 4 Overview of the JPEG Still Picture Compression Algorithm Version 2 0 DCT Based Compression Encoder 8x8 Blocks HH D I Ly Quantizer gt Entropy A Encoder Header Data Source Image Data Quantization Huffman Tables Compressed Tables Image Data Figure 2 1 DCT Based Encoder Functional Blocks The 64 image samples are the first transformed into 64 DCT coefficients by the Forward DCT FDCT The DCT is a relative of the Discrete Fourier Trans form and the FDCT may be regarded intuitively as a harmonic analyzer which decomposes each 8x8 block of pixels into a set of 64 two dimensional 2D spatial frequencies each with a vertical and horizontal component The coef ficient values may then be regarded as a
15. specifies the input file name which may contain a path and must have an extension of JPG dataFrame Specifie typede s an image in system memory as below f struct DATAFRAME BYTE HUGE buffer UWORD dx UWORD dy BYTE UpDownFlip BYTE PixelFormat BYTE Reserved DataFrame Ej Bj Dj The members of the structure are explained below buffer Point to a byte buffer that contains the image data The image size and format are specified by the other members of this structure The buffer length must be calculated as if PixelFormat DATA 8bit BW bp 1 else if PixelFormat DATA 16bit YCbCr422 PixelFormat DATA 16bit YONLY bp 2 length dx dy bp JPEG Software Development Kit User Guide Function References 3 19 E dx Specifies the number of pixels per line in the Data Frame B dy Specifies the number of lines in the Data Frame E UpDownFlip This flag specifies the direction of compressing inverse vertically if set to 1 See the example code for MTRGB24BitToY CbCr422 function on how to use this flag B PixelFormat Specifies the pixel format for the Data Frame Only the following formats are supported DATA 8bit BW 8 bit per pixel Black White format DATA l16bit YCbCr422 16 bit per pixel YCbCr422 format DATA l6bit YONLY 16 bit per pixel YCbCr422 format But only the Y component is relavent in this case the Cb and Cr component will be filled
16. the first line to be compressed is from the top or bottom of the ROI This is useful when the user need to invert the image while it is compressed See the example code for MTRGB2ABitToY CbCr42X function for how to use this flag For lossy baseline compression the Quality Factor is used to control the compressed image quality The higher the Quality Factor the better the compressed image quality and the lower the compression ratio and vice versa To adjust the compression ratio and compressed image quality MuTech recommends using 128 as the start value Usually this value will generate acceptable results for both compression ratio and compressed image quality The user may experiment with different settings and choose the best one For lossless compression a Predictor Number is used For details about the definition of the Predictor Number refer to Appendix C SIDE EFFECT None SEE ALSO JPEG File2Memory Decompress TInquireJPGVals TRGB24BitToYCbCr422 EXAMPLE cCompressParameter cPara DataFrame dFrame Version 2 0 JPEG Software Development Kit User Guide Function References 3 17 ROI roi BYTE monoInfo bpp fill out the input data structure the following code segment is equivalent to dFrame PixelFormat DATA AS COMPRESSED MTInquireJPGVals test jpg amp dFrame dx amp dFrame dy amp colorInfo NULL NULL if monoInfo 0 dFrame PixelFormat
17. with a Data Frame in order to be completely defined The region is used to put the decompressed image into an ROI of a larger Data Frame To decompress the image into entire Data Frame define the region equal to the frame If the ROI is smaller than the image size only the upper left part which is equal to region will be filled into the Data Frame Comments Pointer to a byte buffer that will be used to hold the JPEG comment information if one exists in the compressed code stream A NULL pointer could be used if the user has no interest of the comment content at the time of calling If a buffer pointer is specified its length must be specified by cLength see below for detailed description It is the caller s responsibility to allocate enough space for the buffer otherwise the extra part of the Comments field will be discarded The user can also retrieve the comment information by MTInquireJPGVals function cLength Pointer to an UWORD variable that can be used in a number of ways described below E points to NULL the user is not interested in any information in the comment field The function will not look for Comment Marker and totally ignore the Comments buffer E points to a variable with the value 0 The user is interested in only the length of the comment field If there is no Comment Marker found the return value will be 0 unchanged Otherwise this function will return the actual length of the comment field in the
18. C MAK The following files are also included in the MuTech JPEG SDK but are ex pected to be used with the IV AXX SDK for DOS X MTJPEGNSAMPLESNIVAJPEGNDOSN DJPEG C DJPEG MAK COURF FON Please refer to README DOS for information on how to build and run the samples 2 2 Windows 16 Bit DLL Version 2 0 The WINI6 version of the MuTech JPEG SDK provides a 16 bit Dynamic Link Library DLL MTJPEG16 DLL and an import library MTJPEGI6 LIB which are based on Microsoft Visual C C 1 5x and which work under both Windows 3 1x and Windows 9X Since MTJPEG16 DLL is a standard Windows DLL is should work with other compilers as well Once installed the following files can be found in the sub directories X MTJPEG INC MTJPEG H PORTABLE H STD DEF H X MTUPEG LIB MTJPEG16 LIB MTJPEG16 DLL X MTJPEGNDOCNREADME W16 X MMTJPEGNSAMPLESNJPEGNWINI16N WMEMORY C WMEMORY H WMEMORY RC WMEMORY DEF WMEMORY MAK The following files are also included in the MuTech JPEG SDK but are ex pected to be used with the IV 4XX SDK for Windows 3 1x X MTJPEG SAMPLES IV4JPEG WIN31 WJPEG C WJPEG H WJPEG RC JPEG Software Development Kit User s Guide Using MuTech JPEG SDK 2 3 WJPEG DEF WJPEG MAK Please refer to README W 16 for information on how to build and run the samples 2 3 Windows 32 Bit DLL The WIN32 version of the MuTech
19. Cr422 16 bit per pixel YCbCr422 format DATA l6bit YONLY 16 bit per pixel YCbCr422 format But only the Y component is relavent in this case the Cb and Cr component are treated as dummies region Specifies a Region Of Interest by the ROI structure typedef struct ROI UWORD sx UWORD sy UWORD dx UWORD dy ROI The members of the structure are explained below E sx sy Specify in term of pixels the offset of the upper left corner of the region in a Data Frame E dx dy Specify in term of pixels and lines the width and height of the region in a Data Frame Note a region must be assoicated with a Data Frame in order to be completely defined Only the image in the region will be compressed To compress the entire Data Frame define the region equal to the frame cmpPara Specifies the compression parameters in the CompressParameter structure defined below typedef struct COMPRESSPARAMETER BYTE CompressMethod BYTE QualityFactor BYTE CompressedDataFormat BYTE FAR Comments UWORD CommentLength CompressParamter JPEG Software Development Kit User Guide Function References 3 15 The member of the structure are explained below E CompressMethod Specifies the compression method which can be one of the following pre defined macros JPEG_LOSSY Uses the JPEG Baseline Compression JPEG_LOSSLESS Uses the JPEG Lossless Compression B QualityFactor Specifies a Quality
20. Factor which ranges from 1 to 255 when JPEG_LOSSY is used When JPEG_LOSSLESS is used it selects one of the predictors to be used in compression The valid value is from 1 to 7 For details about predictor refer to Appendix C B CompressedDataFormat Specifies the Compressed Data Format with one of the following pre defined macros JPEG_8bit_BW Specifies that the Compressed Data Format will be 8 bit per pixel Black White format regardless what the Pixel Format is selected in the Data Frame If the Pixel Format in the Data Frame is DATA_16bit_YCbCr or DATA_16bit_YONLY the image data will be converted to DATA_8bit_BW before compression JPEG_16b1t_YCbCr422 Specifies that the Compressed Data Format will be 16 bit per pixel YCbCr422 format This parameter is valid only when the input data format is DATA_16bit_YCbCr422 JPEG_AS SOURCE The Compressed Data Format will be determined by input data i e Pixel Format in the Data Frame When the Pixel Format is DATA_8bit_BW the Compressed Data Format will be JPEG_8bit_BW When the Pixel Format is DATA_16bit_YCbCr422 or DATA_16bit_YONLY the Compressed Data Format will be JPEG_16bit_YCbCr422 E Comments Pointer to a byte buffer that contains application related information such as comment time stamp etc The contents in the buffer will be copied to the compressed file using a JPEG Comment Marker The user can retrieve this information by calling either the MTInquireJPG Vals or
21. Frame B dx dy Specify in term of pixels and lines the width and height of the region in a Data Frame Note a region must be assoicated with a Data Frame in order to be completely defined Only the image in the region will be compressed To compress the entire Data Frame define the region equal to the frame cmpPara Specifies the compression parameters in the CompressParameter structure defined below typedef struct COMPRESSPARAMETER BYTE CompressMethod BYTE QualityFactor BYTE CompressedDataFormat BYTE FAR Comments UWORD CommentLength CompressParamter The member of the structure are explained below B CompressMethod Specifies the compression method which can be one of the following pre defined macros JPEG LOSSY Uses the JPEG Baseline Compression JPEG LOSSLESS Uses the JPEG Lossless Compression B QualityFactor Specifies a Quality Factor which ranges from 1 to 255 when JPEG_LOSSY is used When JPEG_LOSSLESS is used it selects one of the predictors to be used in compression The valid value is from 1 to 7 Foe details about predictor refer to Appendix C JPEG Software Development Kit User Guide Version 2 0 3 24 Function References B CompressedDataFormat Specifies the Compressed Data Format with one of the following pre defined macros JPEG_8bit_BW Specifies that the Compressed Data Format will be 8 bit per pixel Black White format regardless what the Pixel
22. G Memory2Memory Compress 3 21 3 1 10 JPEG_Memory2Memory_Decompress lel 3 27 3 2 MuTech Products Support Group 3 30 3 2 1 IV 4XX Board Support Group 3 30 S321 IMAJPEGLOad ite nme ere ns 3 30 3 2 2 IVAJPEGS Ve 5 ie ete eee eet diee eR 3 34 3 2 1 3 V4InquireJPEGVals eere te redes 3 37 3 3 Misellanegus Utility Group d ee NM lw wax ca we ee x ES 3 39 3 3 1 MTRGB24BitToYCbCr422 o o a ET nee 3 39 3 3 2 MTYCbCr422ToRGB24Bit cui aeons Rr dene pec ies 3 40 3 3 3 MIROGORBiibBtovchC i ene 3 41 3 3 4 MTYCbCr422ToRGB32Bit 0 teens 3 42 3 35 MTInquireJPGV als cin te Bain ii 3 43 JPEG Software Development kit User s Guide Version 2 0 2 Table of Contents 3 3 6 MTInquireJPGInMemoryVals eee 3 44 3 3 7 MIInquimeBMPVals beet bes RR PS PR des 3 46 3 3 48 MTInquireTGAVals seeeeeeeeeee I eens 3 47 3 3 9 MTInquireTIFFVals o e i on ai ea n III 3 48 A Glossary B Overview of the J PEG Still Picture Compression Algorithm B 1 B 2 B 3 B 4 B 5 B 6 B 7 B 8 Introduction B 1 Algorithm Requirements and Selection Process B 2 Basic DCT Block Diagram B 3 Source Images and Data Interleaving B 6 Sequential Build up and Baseline System B 7 Progressive Build up B 7 Hierarchical Encoding B 8 Standa
23. GE buffer UWORD dx UWORD dy BYTE UpDownFlip BYTE PixelFormat BYTE Reserved DataFrame w p E The members of the structure are explained below region buffer Point to a byte buffer that contains the image data The image size and format are specified by the other members of this structure The buffer length must be calculated as if PixelFormat DATA 8bit BW bp 1 else if PixelFormat DATA 16bit YCbCr422 PixelFormat DATA 16bit YONLY bp 2 length dx dy bp dx Specifies the number of pixels per line in the Data Frame dy Specifies the number of lines in the Data Frame UpDownFlip This flag specifies the direction of compressing inverse vertically if set to 1 PixelFormat Specifies the pixel format for the Data Frame Only the following formats are supported DATA 8bit BW 8 bit per pixel Black White format DATA l16bit YCbCr422 16 bit per pixel YCbCr422 format DATA l6bit YONLY 16 bit per pixel YCbCr422 format But only the Y component is relavent in this case the Cb and Cr component are treated as dummies Specifies a Region Of Interest by the ROI structure JPEG Software Development kit User Guide Function References 3 23 typedef struct ROI UWORD sx UWORD sy UWORD dx UWORD dy ROI The members of the structure are explained below E sx sy Specify in term of pixels the offset of the upper left corner of the region in a Data
24. PTIO N PARAM ETERS RETURN VALUE COMMENTS SEE ALSO include mtjpeg h void MTPROCALL MTRGB32BitToYCbCr422 void HUGE input void HUGE output ULONG length UWORD RGBOrder UWORD SubSample Converts a stream of RGB data to a stream of YCbCr422 YUY V data input Pointer to the input buffer which contains RGB data to be converted The length of the input buffer must be at least 4 times the number of pixels length 4 The data sequence of the input buffer is defined by the RGBOrder parmeter output Pointer to the output buffer which is used to store YCbCr The length of the output buffer must be at least 2 times the number of pixels length 2 The data format of the output buffer is YCbCr422 YUYV where each unit corresponds to each pair of RGB pixels length Specifies the number of pixels to be converted The length must be even number RGBOrder Specifies the order of the RGB data strored in the input buffer When RGBOrder is 1 the input data format is RGBXRGBX when RGBOrder is O the input data format is BGRXBGRX where X is the byte discarded in the conversion SubSample Reserved and must be set to 0 None Also see comments for JPEG_Memory2File_Compress MTYCbCr422RGB32Bit JPEG Memory2File Compress JPEG Software Development kit User Guide Version 2 0 3 42 Function References 3 3 4 MTYCbCr422ToRGB32Bit SYNOPSIS DESC RIPTIO N PARAMETERS RETURN
25. RESSED test tga NULL NULL Les OK JPEG Software Development Kit User Guide Function References 3 11 error occurs release buffer MTFREE buffer 3 1 6 JPEG_Memory2File_Decompress SYNOPSIS DESC RIPTIO N PARAMETERS include mtjpeg h int MTPROCALL JPEG Memory2File Decompress BYTE HUGE jpegBuffer ULONG bufferLength BYTE PixelFormat char FAR OutFile BYTE FAR Comments UWORD FAR cLength Decompresses a JPG format memory buffer into a BMP TGA or TIF file jpegBuffer A pointer to a BYTE buffer that holds the compressed code stream in JPG format bufferLength Specifies the length of the buffer The bufferLength defines the byte number of code in the buffer The bufferLength usually has a value returned from previous call to JPEG_File2Memory_Compress function PixelFormat This parameter must be filled with one of the pre defined macros DATA AS COMPRESSED Used for most cases see comments for details DATA l16bit YCbCr422 This is an exception that may be used only when the compression was lossy and the OutFile has an extension of TIF See comments for details OutFile A pointer to a string that specifies the output file name which may contain a path and must have an extension of BMP TGA or TIF lt Comments Pointer to a byte buffer that will be used to hold the JPEG comment information if one exists in the compres
26. SIS DESC RIPTIO N PARAMETERS include mtjpeg h int MTPROCALL JPEG Memory2Memory Decompress BYTE HUGE jpegBuffer ULONG bufferLength DataFrame dataFrame ROI region BYTE FAR Comments UWORD FAR cLength Decompresses a JPG format image stored in a memory buffer into a pre defined Data Frame in system memory jpegBuffer A pointer to a BYTE buffer that holds the compressed code stream in JPG format bufferLength Specifies the length of the buffer The bufferLength defines the byte number of code in the buffer The bufferLength usually has a value returned from previous call to JPEG Memory2Memory Compress function dataFrame Specifies an image in system memory as below typedef struct DATAFRAME BYTE HUGE buffer UWORD dx UWORD dy BYTE UpDownFlip BYTE PixelFormat BYTE Reserved DataFrame The members of the structure are explained below Ed p E B buffer Point to a byte buffer that contains the image data The image size and format are specified by the other members of this structure The buffer length must be calculated as if PixelFormat DATA 8bit BW bp 1 else if PixelFormat DATA 16bit YCbCr422 PixelFormat DATA 16bit YONLY bp 2 length dx dy bp B dx Specifies the number of pixels per line in the Data Frame JPEG Software Development kit User Guide Version 2 0 3 28 Function References dy Specifies the n
27. Version 2 0 JPEG Software Development kit User s Guide Using MuTech JPEG SDK 2 1 2 Usng Mulechj PEG SDK The MuTech JPEG SDK is provided separately for different platforms such as DOS WIN16 and WIN32 SDK s for other platforms may also be available Call MuTech Corporation for information The MuTech JPEG SDK adds JPEG compression decompression capabilities to other MuTech products It currently supports the IV 4XX product line to gether with the IV AX X SDK Support for other MuTech products will become available in the future Generally the MuTech JPEG SDK is a software pack age which can be used either alone or with any other MuTech SDKs The following sections describe the contents and the organizations of the JPEG SDK for different platforms 2 1 DOS 16 Bit Library The DOS version of the MuTech JPEG SDK provides a group of 16 bit librar ies MTJPEG LIB FIOLIB LIB and IV4JPEG LIB which are based on Mi crosoft Visual C C 1 5x and work only with the MSC compiler Only large memory mode libraries are provided Once installed the following files can be found in the sub directories X MTJPEG INC MTJPEG H PORTABLE STD DEF H H X MMTJPEGNLIBN MTJPEG LIB FIOLIB LIB IVAJPEG LIB X MTJPEGNDOCNREADME DOS X MTJPEGNSAMPLESNJPEGNDOSN JPEG Software Development kit User s Guide Version 2 0 2 2 Using MuTech JPEG SDK DF2FCMP C DF2FCMP MAK DF2FDEC C DF2FDE
28. _File2Memory_Compress 3 1 7 JPEG_Memory2File_Compress SYNOPSIS tinclude mtjpeg h int MTPROCALL JPEG Memory2File Compress DataFrame dataFrame ROI region CompressParameter cmpPara BYTE FAR OutFile DESCRIPTION Compresses an image from a pre defined Data Frame in system memory into a JPG file PARAMETERS dataFrame Specifies an image in system memory as below typedef struct DATAFRAM BYTE HUGE buffer UWORD dx UWORD dy BYTE UpDownFlip BYTE PixelFormat BYTE Reserved Ed Bj Dj DataFrame The members of the structure are explained below E buffer Point to a byte buffer that contains the image data The image size and format are specified by the other members of this structure The buffer length must be calculated as if PixelFormat DATA 8bit BW bp 1 else if PixelFormat DATA 16bit YCbCr422 PixelFormat DATA 16bit YONLY bp 2 length dx dy bp dx Specifies the number of pixels per line in the Data Frame dy Specifies the number of lines in the Data Frame E UpDownFlip JPEG Software Development Kit User Guide Version 2 0 3 14 Function References Version 2 0 This flag specifies the direction of compressing inverse vertically if set to 1 B PixelFormat Specifies the pixel format for the Data Frame Only the following formats are supported DATA 8bit BW 8 bit per pixel Black White format DATA l16bit YCb
29. _SUPPORTED The IV 4XX SDK could not find the MuTech JPEG SDK DLL file This error code is for Windows 3 1x 9x NT only IV4_PARAM_ERR Indicates one of the following cases E sx gt the width of the IV 4XX Video Frame WB sy gt the height of the IV 4XX Video Frame B name does not have an extension of JPG invalid flag COMMENTS For lossy baseline compression Quality Factor is used to control the compressed image quality The higher the Quality Factor the better the compressed image quality and the lower the compression ratio and vice versa To adjust the compression ratio and compressed image quality MuTech recommends using 128 as the Quality Factor to start value Usually this value will generate acceptable results for both compression ratio and compressed image quality The user may experiment with different settings and choose the best one For lossless compression a Predictor Number is used For details about the definition for the Predictor Number refer to Appendix C SEE ALSO IV4SetVideoFrame IVAJPEGLoad EXAMPLE UWORD DefaultSx DefaultSy this example grab an single frame into the Nideo Frame and compress and save to a JPG file open and initialize the board set to use NTSC camera IV4SetTVStandard IVA Camera NTSC the default sx sy may vary among different TV standards and also among different cameras DefaultSx 0x0033 DefaultSy 0x000A set grab window to grab fram
30. ation inquiry capability 3 3 1 MTRGB24BitToY CbCr422 SYNOPSIS DESC RIPTIO N PARAM ETERS RETURN VALUE COMMENTS SEE ALSO include mtjpeg h void MTPROCALL MTRGB24BitToYCbCr422 void HUGE input void HUGE output ULONG length UWORD RGBOrder UWORD SubSample Converts a stream of RGB data to a stream of YCbCr422 YUYV data input Pointer to the input buffer which contains the RGB data to be converted The length of the input buffer must be at least 3 times the number of pixels length 3 The data sequence of the input buffer is defined by the RGBOrder parmeter output Pointer to the output buffer which is used to store the resulte of the conversion The length of the output buffer must be at least 2 times the number of pixels length 2 The data format of the output buffer is YCbCr422 YUYV where each unit corresponds to each pair of RGB pixels length Specifies the number of pixels to be converted The length must be even number RGBOrder Specifies the order of the RGB data strored in the input buffer When RGBOrder is 1 the input data format is RGBRGB when RGBOrder is 0 the input data format is BGRBGR SubSample Reserved and must be set to 0 None Also see comments for JPEG_Memory2File_Compress MTYCbCr422ToRGB24Bit JPEG Memory2File Compress JPEG Software Development kit User Guide Version 2 0 3 40 Function References 3 3 2 MTYCbCr422ToRGB24B
31. bit YCbCr422 IV4_Load_As_BW Black White or Color 8 bit B W Cb Cr 0x80 JPEG Software Development Kit User Guide Version 2 0 3 32 Function References RETURN VALUE Version 2 0 Comments Pointer to a byte buffer that will be used to hold the JPEG comment information if one exists in the compressed code stream A NULL pointer could be used if the user has no interest of the comment content at the time of calling If a buffer pointer is specified its length must be specified by cLength see below for detailed description It is the caller s responsibility to allocate enough space for the buffer otherwise the extra part of the Comments field will be discarded The user can also retrieve the comment information by calling MTInquireJPGVals function cLength Pointer to an UWORD variable that can be used in a number of ways described below E points to NULL the user is not interested in any information in the comment field The function will not look for Comment Marker and totally ignore the Comments buffer E points to a variable with the value 0 The user is interested in only the length of the comment field If there is no Comment Marker found the return value will be 0 unchanged Otherwise this function will return the actual length of the comment field in the variable pointed to The Comments buffer will be ignored E points to a variable with non zero value The user is interested in both the len
32. ci cta Ree a ER hg 3 5 V version number 3 2 W WIN32 eet Beane es ba eb A AS 2 3 Windows OX nees sce i digs o a 2 3 Windows NT ii is Baie 2 3 Y YCbGr 5 ee tht 1 2 3 2 3 42 YCbCr422 3 5 3 8 3 14 3 16 3 19 3 21 3 31 3 39 3 43 A EE 3 39 JPEG Software Development kit User s Guide
33. comment field in the variable pointed to OK The function completed successfully PARA ERROR One of the parameters is invalid The Data Frame structure except the buffer must be filled out accordingly before passed to this function The decompressed image data will be stored in the frame buffer one line adjacent to another without gaps The length of a JPEG Software Development Kit User Guide SIDE EFFECT SEE ALSO EXAM PLE Function References 3 21 line in bytes depends on PixelFormat 8 bit Black White format takes 1 byte per pixel 16 bit YCbCr422 format takes 2 bytes per pixel The total length of a line is the number of pixels per line dx times the number of bytes per pixel How to determine the PixelFormat migth be tricky because the result will also depends on the Compressed Data Format in the input file The easiest way is to use DATA AS COMPRESSED It will determine the PixelFormat based on the Compressed Data Format in the input file See parameter description above for DATA AS COMPRESSED This function has some ability to convert data before writting to the buffer For example assuming the Compressed Data Format is YCbCr422 color image If the user specifies the PixelFormat as DATA 8bit BW the Cb and Cr components will be dropped and the image will be packed to 8 bit Black White format If the user specifies the PixelFormat as DATA 16bit YONLY the Cb U and Cr V components will be replaced with 0x80 Othe
34. d in compression The valid value is from 1 to 7 B CompressedDataFormat This entry is reseved and must be filled with the pre defined macro JPEG AS SOURCE See comments for details B Comments Pointer to a byte buffer that contains application related information such as comment time stamp etc The contents in the buffer will be copied to the compressed jpegBuffer using a JPEG Comment Marker The user can retrieve this information by calling either the MTInquireJPGInMemory Vals or one of the decompression functions that decompresses from a JPG memory buffer E CommentLength Specifies the length of the information in bytes If the contents is longer than CommentLength the extra part will be dropped The user must ensure that the CommentLength is not longer than the size of the Comments buffer E jpegBuffer A pointer to a buffer of bytes which will be used to store the compressed code stream in JPG format bufferLength A ULONG variable that specifies the length of the buffer The requirement of the buffer length depends on the compression ratio which is in turn depending on the Quality Factor parameter It is the caller s responsibility to allocate enough space to accommodating the code stream A rule of thumb is to allocate a buffer the same size of the original image RETURN VALUE OK The function completed successfully JPEG Software Development Kit User Guide Version 2 0 3 10 Function References COMMENTS
35. de produced by the Huffman encoder An entropy encoding procedure which assigns a variable length code to each input symbol The set of variable length codes required for Huffman encoding and Huffman decoding The Joint Photographic Experts Group This group is a joint ISO CCITT tech nical committee ISO IEC JTC1 SC2 WG10 Photographic Image Coding whose goal has been to develop a general purpose international standard for gray scale or true color compression A term for encoding and decoding processes and procedures in which the out put of the decoding procedure s is identical to the input to the encoding pro cedure s A term for encoding and decoding processes which are not lossless Version 2 0 A 2 Glossary predictor quantization 8x8 block Sub Sa mpling Version 2 0 A linear combination of previously encoded reconstructed values uses in loss less mode coding The procedure by which the DCT coefficients are linearly scaled in order to achieve better compression An 8x8 array of samples A procedure by which the spatial resolution of an image is reduced JPEG Software Development Kit User s Guide Overview of the JPEG Still Picture Compression Algorithm B 1 AppendixB Overview of the J PEG Still Picture Compression Algonthm By Gregory K Wallace Digital Equipment Corporation Maynard Massachusetts B 1 Introduction For the past few years a standardization effort known by the acro
36. ding applications The lossless procedures tend to produce about 2 1 compression for color imag es with moderately complex scenes JPEG Software Development kit User s Guide Version 2 0 B 6 Overview of the JPEG Still Picture Compression Algorithm B 4 Source Imagesand Data Interleaving Version 2 0 The previous section described how a single component of an image is decom posed into 8x8 blocks and then transformed quantized and coded for the DCT based modes of operation While some images grayscale or mono chrome images consist only of a single component the JPEG standard is de signed to handle also the more general class of multiple component images which includes all varieties of color images and the other types as well A multiple component image maybe compressed in either interleaved or non interleaved order In the latter case each image component is compressed in its entirety and output as part of the compressed datastream before starting the next component In the interleaved case 8x8 blocks from each component are processed in round robin fashion and compressed blocks are output in in terleaved order into the compressed stream The case of a three component image being encoded in interleaved order is shown in Figure B 2 For sim plicity only one set of tables is shown Only one table can be assigned to each component so the encoder must switch to the appropriate table as it switches between source i
37. e as the orig inal The degree of distortion lt N gt is inversely proportional to the Quality Fac tor spesified during compression The higher the Quality Factor the less the image distortion Please refer to Appendix B for details on Baseline compres sion and decompression 1 3 Lossless Compression and Decompres sion Lossless Codec Compression and Decompression is less complicated than Baseline Codec The image is not divided into blocks and there is no DCT transformation quantization or color space conversion JPEG uses the original image s color components for compression and decompression The compression is based on neighborhood prediction and Huffman coding Block diagrams for lossless compression and decompression are shown in Fig ure 1 4 and Figure 1 5 respectively Compression Lossless Codec uses a predictive coding technique based on the Differential Pulse Code Modulation DPCM model The prediction error from the DPCM is encoded by the same Huffman coding used by Baseline codec The follow ing are the steps for Lossless compression Figure 1 4 1 A predictor combines the reconstructed values of up to three neighbor hood samples at positions a b and c to form a prediction of the sample at position x as shown in Figure 1 3 JPEG Software Development kit User s Guide Version 2 0 1 4 Introduction 2 This prediction is subtracted from the actual value of the sample at the position x generating the d
38. e with odd field as first field IV4SetGrabWindow IV4 Frame Mode IV4_Grab_Odd DefaultSx DefaultSy IVA NTSC Width Default 640 IV4 NTSC Height Default 480 set the Video Frame IV4SetVideoFrame IV4 Single Image 0 0 IVA NTSC Width Default 640 IV4 NTSC Height Default 480 Version 2 0 JPEG Software Development Kit User Guide Function References 3 37 grab one frame into the Video Frame IV4StartGrab IV4_Single_Grab the following call will save an image from the Video Frame to a JPG file IV4JPEGSave image jpg 0 O IV4_NTSC_Width_Default 640 IV4_NTSC_Height_Default 480 IVA Save As Is IV4_JPEG_LOSSY 128 NULL 0 3 2 1 3 V4InquireJPEG Vals SYNOPSIS DESC RIPTIO N PARAMETERS include iv4 h int MTPROCALL IV4InquireJPEGVals char name UWORD dx UWORD dy UWORD fmt BYTE FAR Comments UWORD FAR cLength Inquires the image size and format in a specified JPG file name specifies the file name The file name can contain a path and must have an extension of JPG dx dy pointer to UWORD variables that return the size of the image in terms of pixels and lines in the JPG file fmt pointer to an UWORD variable that returns the color depth of the image in JPG file The possible values returned can be 8 or 16 bits Comments Pointer to a byte buffer that will be used to hold the JPEG comment information if
39. ecompression function JPEG File2File Decompress get code stream size CodeSize JPEG InquireCodeStreamSize now CodeSize contains the number of bytes from the input file that has been decoded JPEG Software Development Kit User Guide Version 2 0 3 4 Function References 3 1 3 JPEG File2File Compress SYNOPSIS DESC RIPTIO N PARAM ETERS Version 2 0 include mtjpeg h int MTPROCALL JPEG File2File Compressq char FAR InFile CompressParameter cmpPara char FAR OutFile Compresses an image file in BMP TGA or TIF format into a JPG file InFile A pointer to a string that specifies the input file name which may contain a path and must have an extension of BMP TGA or TIF cmpPara Specifies the compression parameters in the CompressParameter structure defined below typedef struct COMPRESSPARAMETER BYTE CompressMethod BYTE QualityFactor BYTE CompressedDataFormat BYTE FAR Comments UWORD CommentLength CompressParamter d PI Pi E The member of the structure are explained below B CompressMethod Specifies the compression method which can be one of the following pre defined macros JPEG_LOSSY Uses the JPEG Baseline Compression JPEG_LOSSLESS Uses the JPEG Lossless Compression B QualityFactor Specifies a Quality Factor which ranges from 1 to 255 when JPEG_LOSSY is used When JPEG_LOSSLESS is used it selects one of the predic
40. ed on a blind assessment of subjective picture quality and narrowed 12 proposed algorithms to three Three informal working groups formed to refine them and in January 1988 a second similar selection process revealed that the ADCT proposal based on the 8x8 Discrete Cosine Transform DCT had produced the best picture qual ity Since then an intensive group wide effort to refine test and document the DCT based algorithm in all its modes of operation has been underway Details of this selection and evaluation process as well has history of the JPEG committee are contained in 1 2 3 4 The requirement for a lossless mode of operation proved to be the most diffi cult one to satisfy with a DCT based algorithm For some time JPEG consid ered the approach whereby the encoder would reconstruct the same compressed data sent to the decoder form the difference of the reconstructed and source images and send the difference image to the decoder But this method does not guarantee a lossless result unless both encoder and decoder use identical Inverse DCT DCT implementations The JPEG committee was reluctant to specify some unique IDCT as a require ment in its proposed standard because research into new fast DCT trans formed algorithm continues and no one algorithm is optimal for all implementations Instead the committee chose to specify a simple predictive method which is entirely separate and independent from the DCT based algo
41. efer to README W32 for information on how to build and run the samples Version 2 0 JPEG Software Development Kit User s Guide Function References 3 1 3 Function References The APIs documented in this chapter have identical interfaces for all platforms supported The functions in Mu Tech JPEG SDK have been divided into several groups the Application Level Group the MuTech Product Support Group and the Miscellanous Utility Group Each will be discussed in this chapter Application Level Group provides easy to use interfaces for most of the applications to compress decompress images between files and the system memory This group of functions can be used in any applications that may or may not relate to Mu Iech products Mu Tech Product Support Group includes functions which must be used with other Mu Tech products Generally these routines will perform compression or decompression between Video Buffer and disk files Currently only IV 4XX product line are supported Supports for other product line may be available in the future Miscellaneous Utility Group provides miscellaneous supports to applications such as RGB and YCbCr color space conversion or file information inquire functions There is a lower level function group called the Developer Level Group in the JPEG SDK This group is the base for all higher level function groups e g the Application Level and the MuTech Product Support groups To use the routines in the Deve
42. esent the major version number and the lower 8 bits represent the minor version number None UWORD SDKVersion inquires SDK version SDKVersion JPEG InquireSDKVersion reports SDK version printf SDK Version is x x SDKVersion gt gt 8 SDKVersion amp 0x00ff JPEG Software Development kit User Guide Function References 3 3 3 1 2 JPEG_InquireCodeStreamSize SYNOPSIS include mtjpeg h ULONG MTPROCALL JPEG InquireCodeStreamSize DESCRIPTION Returns the total number of bytes of the code stream processed by the last compression decompression function call PARAMETERS None RETURN VALUE A 32 bit integer vaule which is the number of bytes of the code stream processed by the last compression decompression function call COMMENTS This function can be called following any one of the compression decompression function calls It will return the code stream size processed by the last compression decompression function Or if none of the compression decompress functions has been called this function should return a value 0 The code stream size can be used to calculate the compression ratio SIDE EFFECT None EXAMPLE ULONG CodeSize called a compression function JPEG File2File Compress get code stream size CodeSize JPEG InquireCodeStreamSize now CodeSize contains the number of bytes of the compressed file another example called a d
43. fer in JPG format o fmt Pointer to an UWORD variable that returns the image type The possible values returned are 0 for color image YCbCr422 or 1 for Black White image 8 bit B W E Comments Pointer to a byte buffer that will be used to hold the JPEG comment information if one exists in the compressed code stream A NULL pointer could be used if the user has no interest of the comment content at the time of calling If a buffer pointer is specified its length must be specified by cLength see below for detailed description It is the caller s responsibility to allocate enough space for the buffer otherwise the extra part of the Comments field will be discarded The user can also retrieve the comment information by using any one of the decompression functions that decompresses from a memory buffer in JPG file format E cLength Pointer to an UWORD variable that can be used in a number of ways described below E points to NULL the user is not interested in any information in the comment field The function will not look for Comment Marker and totally ignore the Comments buffer E points to a variable with the value 0 The user is interested in only the length of the comment field If there is no Comment Marker found the return value will be 0 unchanged Otherwise this function will return the actual length of the comment field in the variable pointed to The Comments buffer will be ignored E points to a variable with non
44. frequency spectrum for that 8x8 block Secondly the coefficients within the current band need not be encoded to their full quantized accuracy within each scan Upon a coefficient s first encod ing the N most significant bits can be encoded first where N is specifiable In one or more subsequent refinement scans the bits of less significance can then be encoded This second mode is called successive approximation Hierarchical Encoding Version 2 0 The hierarchical mode 2 provides a pyramidal encoding of an image at multiple resolutions each differing in resolution from its adjacent encoding by a factor of two in either the horizontal or vertical dimension or both The en coding procedure can be summarized as follows a Filter and down sample the original image by the desired number of multiples of 2 in each dimension b Encode this reduced size image using either the sequential or progres sive build up methods previously described c Decode this reduced size image and then interpolate and up sample it by 2 horizontally and or vertically using the identical interpolation filter which the receiver must use d Use this up sampled image as a prediction of the original at this reso lution and encode the difference image using either the sequential or progres sive encoding modes e Repeat steps c and d until the full resolution of the image has been encoded Hierarchical encoding is useful in app
45. g h int MTPROCALL MTInquireJPGVals char FAR file UWORD FAR dx UWORD FAR dy UWORD FAR fmt BYTE FAR Comments UWORD FAR cLength Inquires the image size and format in a specified JPG file file Specifies the file name The file name can contain a path and must have an extension of JPG dx dy Pointer to UWORD variables that return the size of the image in terms of pixels and lines in the JPG file fmt Pointer to an UWORD variable that returns the image type The possible values returned are 0 for color image YCbCr422 or 1 for Black White image 8 bit B W Comments Pointer to a byte buffer that will be used to hold the JPEG comment information if one exists in the compressed code stream A NULL pointer could be used if the user has no interest of the comment content at the time of calling If a buffer pointer is specified its length must be specified by cLength see below for detailed description It is the caller s responsibility to allocate enough space for the buffer otherwise the extra part of the Comments field will be discarded The user can also retrieve the comment information by using any one of the decompression functions that decompresses from a JPG file E cLength Pointer to an UWORD variable that can be used in a number of ways described below E points to NULL the user is not interested in any information in the comment field The function will not look for Comment Marker and totally i
46. gnore the Comments buffer E points to a variable with the value 0 The user is interested in only the length of the comment field If there is no Comment Marker found the return value will be 0 unchanged Otherwise this function will return the actual length of the comment field in the variable pointed to The Comments buffer will be ignored JPEG Software Development kit User Guide Version 2 0 3 44 Function References RETURN VALUE COMMENTS SEE ALSO EXAM PLE E points to a variable with non zero value The user is interested in both the length and the content of the comment field In this case the Comments buffer must be allocated and supplied and the value of the variable must be equal to the size of the Comments buffer If there is no Comment Marker found the return value will be 0 Otherwise this function will copy the comment field into the buffer pointed by Comments and return the actual length of the comment field in the variable pointed to OK The function completed successfully PARAM ERR file does not have an extension of JPG The application calls this function to get the image size and the data format of the JPG file It is helpful that the application knows the information before further operates on the file In general there two ways that the application can retrieve the comment field from a compressed code stream The first is to call this function to find out the information about the comment f
47. gth and the content of the comment field In this case the Comments buffer must be allocated and supplied and the value of the variable must be equal to the size of the Comments buffer If there is no Comment Marker found the return value will be 0 Otherwise this function will copy the comment field into the buffer pointed by Comments and return the actual length of the comment field in the variable pointed to IVA OK The function completed successfully IVA ERROR The board has not been initialized IV4 NOT SUPPORTED The IV AXX SDK could not find the MuTech JPEG SDK DLL file This error code is for Windows 3 1x 9x NT only IVA PARAM ERR Indicates one of the following cases H sx gt the width of the image in the file E sy gt the height of the image in the file W dsx gt the width of the Video Frame B dsy the height of the Video Frame JPEG Software Development Kit User Guide Function References 3 33 B name does not have an extension of JPG W invalid flag COMMENTS This function requires the application to define a IV AXX Video Frame by calling IV4SetVideoFrame before loading the image The parameter dsx dsy and dx dy are referenced to the Video Frame The application can call IV4InquireJPEGVals to get the image size and the data format about the compressed image and then set the IV AXX Video Frame accordingly In General there two ways that the application can retrieve the comment field fro
48. he user 3 Huffman code tables are then used to encode the entropy of the quan tized DCT coefficients These Huffman code tables are pre defined in side the MuTech JPEG SDK An 8 x 8 Block of Original Image Ly FDCT Quantizer Loy Entropy Encoder Quantization Huffman Table Code Table Compressed Image Data Figure 1 1 Baseline Encoding Diagram Decompression Reversing the compression process an image is restored from the compressed code stream as follows Figure 1 2 1 The compressed code stream is decoded into DCT coefficients by us ing the same Huffman code tables used in the compression process Version 2 0 JPEG Software Development Kit User s Guide Introduction 1 3 2 The DCT coefficients are dequantized using the same quantization ta bles as in the compression process 3 The group of 64 dequantized DCT coefficients are transformed back into 8x8 blocks of pixels using Inverse DCT Se gt Dequantizer gt rper Compressed Image Data A Decompressed Quantization Huffman Image Data Table Code Table Figure 1 2 Baseline Decoding Diagram After the decompression the result will be converted into an RGB color image if the user desires The restored image will not be exactly the sam
49. ield then allocate the buffer of the right size and retrieve the comment field during the decompression function call The second way is to blindly allocate a buffer which is large enough to hold the maximum anticipated size of the comment field then calls this function to get the information needed When one of the decompression function is called using the NULL pointer option to disregard the comment field JPEG Memory2File Compress See the example for JPEG Memory2File Compress 3 3 6 MTInquireJPGInMemory Vals SYNOPSIS DESC RIPTIO N PARAM ETERS Version 2 0 include mtjpeg h int MTPROCALL MTInquireJPGInMemoryVals BYTE HUGE jpegBuffer ULONG bufferLength UWORD FAR dx UWORD FAR dy UWORD FAR fmt BYTE FAR Comments UWORD FAR cLength Inquires the image size and format of compressed data in a specified memory buffer in JPG file format jpegBuffer Pointer to a BYTE buffer that will be used to store the compressed data in JPG format bufferLength JPEG Software Development Kit User Guide Function References 3 45 Specifies the length of the jpegBuffer The bufferLength defines the byte number of the code in the buffer It usually has a value returned from provious call to a comprssion function that is using the memory buffer to hold the compressed code stream dx dy Pointer to UWORD variables that return the size of the image in in terms of pixels and lines in the memory buf
50. ifference 3 The difference is then coded by Huffman coding Figure 1 3 Three sample Prediction Neighborhood Predictor Entropy y Original Prediction Encoder Compressed Image Data Differences Code Stream Huffman Code Table Figure 1 4 Lossless Encoding Diagram Decompression The procedure below restores an image from a compressed code stream Figure 1 5 1 The Huffman code is decoded for the prediction difference of each pixel 2 The prediction value for each pixel is calculated This is based on the same predictor used in the encoding process and the same reconstruct ed neighborhood samples for example a b and c as shown in Figure 1 3 3 To restore the original image the prediction value of each pixel gen erated from Step 2 is added to the prediction difference of each pixel decoded from Step 1 Version 2 0 JPEG Software Development Kit User s Guide Compressed Code Stream Entropy Decoder Prediction Differences Reverse Predictor Huffman Code Table Figure 1 5 Lossless Decoding Diagram Introduction 1 5 Decompressed Image Data After the decompression the restored image will be exactly the same as the original Refer to Appendix C for definitions of different predictors used by JPEG lossless compression JPEG Software Development kit User s Guide Version 2 0 1 6 Introduction
51. it SYNOPSIS DESC RIPTIO N PARAMETERS RETURN VALUE COMMENTS SEE ALSO EXAM PLE Version 2 0 include mtjpeg h void MTPROCALL MTYCbCr422ToRGB24Bit void HUGE input void HUGE output ULONG length UWORD RGBOrder UWORD SubSample Converts a stream of YCbCr data to a stream of RGB data input Pointer to the input buffer which contains the YCbCr data to be converted The length of the input buffer must be at least 2 times the number of pixels length 2 The data is YCbCr422 YUYV where each unit corresponds to each pair of RGB pixels output Pointer to the output buffer which is used to store the results of the conversion The length of the output buffer must be at least 3 times the number of pixels length 3 The data sequence of the output buffer is defined by the RGBOrder parmeter length Specifies the number of pixels to be converted The length must be even number RGBOrder Specifies the order of the RGB data strored in the output buffer When RGBOrder is 1 the output data format is RGBRGB when RGBOrder is 0 the output data format is BGRBGR SubSample Reserved and must be set to 0 None Also see comments for JPEG_File2Memory_Decompress MTRGB24BitToYCbCr422 JPEG File2Memory Decompress See example for MTRGB2ABitToY CbCr42X JPEG Software Development Kit User Guide Function References 3 41 3 3 3 MTRGB32BitToY CbCr422 SYNOPSIS DESC RI
52. ld then calls this function to get the information needed None JPEG File2File Compress MTInquireJPGVals See sample for JPEG File2File Compress 3 1 5 JPEG File2Memory Compress SYNOPSIS DESC RIPTIO N PARAM ETERS Version 2 0 include mtjpeg h int MTPROCALL JPEG File2Memory Compress char FAR InFile CompressParameter cmpPara BYTE HUGE jpegBuffer ULONG FAR bufferLength Compresses an image file in BMP TGA or TIF format into a memory buffer in JPG file format InFile A pointer to a string that specifies the input file name which may contain a path and must have an extension of BMP TGA or TIF cmpPara Specifies the compression parameters in the CompressParameter structure defined below typedef struct COMPRESSPARAMETER BYTE CompressMethod BYTE QualityFactor BYTE CompressedDataFormat BYTE FAR Comments UWORD CommentLength CompressParamter The member of the structure are explained below JPEG Software Development Kit User Guide Function References 3 9 E CompressMethod Specifies the compression method which can be one of the following pre defined macros JPEG_LOSSY Uses the JPEG Baseline Compression JPEG LOSSLESS Uses the JPEG Lossless Compression B QualityFactor Specifies a Quality Factor which ranges from 1 to 255 when JPEG_LOSSY is used When JPEG_LOSSLESS is used it selects one of the predictors to be use
53. lications in which a very high resolution image must be accessed by a lower resolution device which does not have the buffer capacity to reconstruct the image at its full resolution and then scale it down for the lower resolution display An example scanned and compressed at high resolution for a very high quality printer where the image must also be displayed on a low resolution PC video screen JPEG Software Development kit User s Guide Overview of the JPEG Still Picture Compression Algorithm B 9 B 8 Standardization Schedule The JPEG committee has completed revision 8 7 of its technical specifica tion the details of which are now quite stable The sequential modes of oper ation have been validated by exchange of encoded data among committee members to verify that independently produced codecs operate compatibly This validation procedure is currently underway with the progressive hierar chical and lossless modes of operations as well Unless an error is discovered the technical definition is considered to be frozen by the committee The technical specification is now being converted to the Committee Draft CD which is the document in the official ISO standards format to be formally balloted The CD ballot is expected to commence before the end of 1990 Upon successful completion of this ballot expected before mid 1991 the Draft International Standard DIS document will be balloted Successful comple tion of the DIS pha
54. loper Level Group requires using complicated call back schemes This group of functions are not domcumented here 3 1 Application Level Group This function group uses the following naming conventions JPEG Software Development kit User Guide Version 2 0 3 2 Function References Compression Decompression Parameter used to retum data to caller JPEG_Source2Destination_Compress W Source can be File or Memory in which the input image data are located The data format can be 8 bit Black White 16 bit YCrCb YUYV or 24 bit RGB in some cases W Destination can be File or Memory in which the output code stream will be stored in JPEG format JPEG_Source2Destination_Decompress W Source can be File or Memory in which the input code stream is located in JPEG format W Destination can be File or Memory in which the output iamge data will be stored The data format can be 8 bit Black White 16 bit YCbCr YUYV or 24 bit RGB in some cases To distingush any parameter which can be used to return data information to the caller we put a small left arrow in front of the parameter This format is shown below OutputParameter 3 1 1 JPEG InquireSDK Version SYNOPSIS DESC RIPTIO N PARAM ETERS RETURN VALUE SIDE EFFECT EXAMPLE Version 2 0 include mtjpeg h UWORD MTPROCALL JPEG InquireSDKVersion void Returns the current JPEG SDK version None A 16 bit hex value the higher 8 bits repr
55. lue will be 0 Otherwise this function will copy the comment field into the buffer pointed by Comments and return the actual length of the comment field in the variable pointed to RETURN VALUE OK The function completes successfully PARA ERROR One of the parameter is invalid COMMENTS This function is functionally identical to JPEG File2File Decompress The only difference is that this function reads the compressed code stream from a memory buffer instead of from a disk file The decompressed image files generated by these two functions are identical if they have been compressed by the same controlling parameters Please refer to JPEG File2File Deompress for detailed comments In General there two ways that the application can retrieve the comment field from a compressed code stream The first is to call MTInquireJPGInMemory Vals function to find out the information about the comment field then allocate the buffer of the right size and retrieve the comment field during the decompression function call The second way is to blindly allocate a buffer which is large enough to hold the maximum anticipated size of the comment field then calls this function to get the information needed SIDE EFFECT None SEE ALSO JPEG File2Memory Compress Version 2 0 JPEG Software Development Kit User Guide Function References 3 13 JPEG File2File Decompress MTInquireJPGInMemoryVals EXAMPLE See sample for JPEG
56. m a compressed code stream The first is to call MTInquireJPG Vals function to find out the information about the comment field then allocate the buffer of the right size and retrieve the comment field during the decompression function call The second way is to blindly allocate a buffer which is large enough to hold the maximum anticipated size of the comment field then calls this function to get the information needed SEE ALSO IV4SetVideoFrame IV4InquireJPEGVals EXAMPLE UWORD VF dx VF dy fmt this example loads an image from a JPG file to IV 4XX board and display it Assume the iV 4XX board has been opened and initialized successfully inquire the file information IV4InquireJPEGVals image jpg amp VF dx amp VF dy amp fmt NULL NULL set the Video Frame IV4SetVideoFrame IVA4 Single Image Mode NULL VF dx VF dy the following call will load an image from file to the Video Frame and display at 0 0 IV4GetVideoFrame NULL NULL amp VF dx amp VF dy IV4SetImageFrameOnVideoFrame 0 0 VF dx VF dy IV4SetImageWndOnScreen 0 0 IVAJPEGLoad image jpg 0 0 0 O VF dx VF dy IVA Load As Is NULL NULL JPEG Software Development Kit User Guide Version 2 0 3 34 Function References 3 2 1 2 IV4JPEGSave SYNOPSIS DESC RIPTIO N PARAM ETERS Version 2 0 include iv4 h int MTPROCALL IVAJPEGSave char FAR name UWORD sx
57. mage components Whether interleaved or non interleaved all DCT based algorithms encode each image component independently of all others BL Compression Encoder m gt Header Data Cc e l Table 1 Table 2 Figure 2 1 Componen Interleave and Table Switching Because the JPEG standard handles multiple component images in this general way it is applicable to images represented in practically any color system In deed there exist many different color systems new ones are still being invent ed in current research and many are the subject of ongoing standardization efforts which are independent of JPEG s work Consequently the compressed image header in the JPEG standard does not contain any parameter which in dicates the color space of the compressed image To JPEG such information is specific to the application and not required to decompress the image The source image may also have one or more components that are subsampled with respect to its others though there are restrictions on this flexibility JPEG Software Development Kit User s Guide Overview of the JPEG Still Picture Compression Algorithm B 7 B 5 Sequential Build up and Baseline System Sequential build up refers to the encoding mode in which 8x8 pixel blocks are compressed and output from left to right and top to bottom in a single scan per component If the image components may be in
58. me dx roi dy dFrame dy fill out the input parameter structure cPara CompressMethod JPEG LOSSY cPara QulityFactor 128 cPara CompressedDataFormat JPEG AS SOURC cPara Comments NULLI cPara cLength 0 E allocate buffer to hold compressed data in JPG format assume the compression ratio is greater than 1 1 the following buffer length is considered most conservative that assume nothing is compressed length ULONG dFrame dx ULONG dFrame dy compress JPEG Software Development Kit User Guide Version 2 0 3 26 Function References if JPEG Memory2Memory Compress dFrame foi cPara jpegBuffer amp length OK error occurs now the length contains the actual size of code stream in bytes decompress fill out the input data structure MTInquireJPGInMemoryVals jpegBuffer amp dFrame dx amp dFrame dy amp colorInfo NULL NULL if monoInfo 0 dFrame PixelFormat bp 2 DATA_16bit_YCbCr422 else dFrame PixelFormat DATA 8bit BW bp 1 if JPEG_Memory2Memory_Decompress jpegBuffer length roi dFrame NULL NULL OK error occurs the decompressed data can be processed her MTFREE dFrame buffer MTFREE jpegBuffer Version 2 0 JPEG Software Development kit User Guide Function References 3 27 3 1 10 JPEG_Memory2Memory_Decompress SYNOP
59. measure of the amount of each 2D fre quency present in the 8x8 pixel block The coefficient with zero frequency in both dimensions is known as the DC coefficient and the other 63 are known as the AC coefficients Each of the 64 DCT coefficients is next uniformly quantized in conjunction with a 64 element Quantization Table Q Table which must be specified as an input to the encoder Each element may be specified as any integer value from 1 to 255 and represents the quantizer step size for the corresponding DCT coefficient Quantization is performed by dividing each DCT coefficient by its corresponding Q Table element and rounding to the nearest integer Ideally each stepsize should be chosen by the application as the perceptual threshold for the visual contribution of its corresponding cosine basis function as a func tion of the intended source image format display characteristics and viewing distance The degree of precision required to perform these computations depends on the range of the input pixels A property of the 8x8 FDCT is that the non fractional part of the DCT coefficients has a range which is a factor of 8 greater than that of the input pixels Given that the minimum quantizer step size is one integer level the quantized DCT coefficients increase in size by three bits compared to the input pixels and greater precision than this is required to compute the FDCT itself with adequate accuracy The JPEG committee has chosen to allo
60. mparison to the to tal transmission time at the viewer s screen and is refined by successive scans until reaching the level of picture quality that was established by the quantiza tion tables Referring to Figure 1 the progressive build up mode of operation can be un derstood by imagining the addition of a full image buffer memory at the output of the quantizer before the input to the entropy coder The buffer memory must be of sufficient size to store the image as quantized DCT coefficients each of which is 3 bits large than the per component input pixels In progressive build up after each block of DCT coefficients is quantized it is stored in the coefficient buffer memory because only a portion of its content is encoded and output at each scan through the image This contrasts with se quential build up where all coefficients in a block are quantized and then im JPEG Software Development kit User s Guide Version 2 0 B 8 Overview of the JPEG Still Picture Compression Algorithm B 7 mediately entropy coded and output at their full quantized resolution without buffering There are two independent ways by which the quantized coefficients of each block may be partially encoded within a scan First only a specified band i e only some of the 64 coeffiients need be encoded This mode is called spec tral selection because each band typically contains coefficients which occupy a lower or higher part of the spatial
61. n Algorithm B 2 ferent manufacturers will interoperate But the JPEG committee has a strong additional conviction that most of these applications can be satisfied by a com mon general purpose image compression standard This would facilitate ex change of images across application boundaries as these applications become increasingly internetworked Moreover state of the art methods for continu ous tone image compression often require VLSI implementation to achieve the coding or decoding speed required by many applications a common gener al purpose method would promise to reduce significantly the cost of special ized compression hardware The price of this aim for application independence is a proposed standard which has been fairly long in the making and which on first blush appears somewhat complex by virtue of its multiple modes of operation It does how ever feature a core mode known as the Baseline System which is common to all other modes of operation and which is expected to be sufficient in its own right for a number of applications Algonthm Requirements and Selection Process Version 2 0 The JPEG committee s goal has been to develop a method for continuous tone image compression which meets the following requirements a Be at or near state of the art with regard to compression rate and ac companying image fidelity over a wide range of image quality ratings and es pecially in the range where visual fidelity t
62. nym JPEG for Joint Photographic Experts Group has been working toward establishing the first international digital image compression standard for between CCITT and ISO JPEG convenes officially as the ISO committee with nomenclature JTC1 SC2 WG10 but operates in close collaboration with CCITT SGVIII The only widely used international digital image compression standard in ex istence today is the bi level compression method used by Group 3 and Group 4 facsimile machines This method differs from the methods being standard ized by the JPEG committee in a number of important ways For one the JPEG methods are applicable only to multi level images and not at all to bi level im ages just the opposite of the existing fax methods Moreover the JPEG stan dard aims to be a general purpose technique for applications as diverse as photo videotex desktop publishing graphic arts color facsimile newspaper wire photo transmission medical systems and many others The Group 3 4 compression methods however were standardized expressly for fax machines though they are being adopted for other applications as well As was the case for commonplace facsimile before the Group 3 standards were established each of these continuous tone imaging applications needs its own image compression standard in order to flourish so that equipment from dif JPEG Software Development kit User s Guide Version 2 0 B 2 Overview of the JPEG Still Picture Compressio
63. o the original is characterized as very good to excellent to indistinguishable b Be applicable to practically any kind of scene content not be limited for example to classes of imagery with restrictions on scene complexity range of colors statistical properties etc c Have tractable computational complexity with some preference given to the decoding side to yield software implementations with viable perfor mance on a range of CPUs as well as hardware implementation with viable cost for applications requiring high performance d Have the following modes of operation W Sequential build up each component of the image is encoded in a sin gle left to right top to bottom scan W Progressive build up the image is encoded in multiple scans for appli cations in which transmission time is long and the user prefers to watch the image build up in multiple coarse to clear passes W Hierarchical encoding the image is encoded at multiple resolutions so that lower resolution versions may be accessed without having first to decompress the image s full resolution JPEG Software Development Kit User s Guide Overview of the JPEG Still Picture Compression Algorithm B 3 W Lossless compression an encoding which will guarantee exact recov ery of every source image pixel value in spite of relatively poor com pared to lossy with excellent fidelity compression rate In June 1987 JPEG conducted a selection process bas
64. one of the decompression functions that decompresses from a JPG file E CommentLength JPEG Software Development Kit User Guide Version 2 0 3 16 Function References Specifies the length of the information in bytes If the contents is longer than CommentLength the extra part will be dropped The user must ensure that the CommentLength is not longer than the size of the Comments buffer OutFile A pointer to a string that specifies the output file name which may contain a path and must have an extension of JPG RETURN VALUE OK The function completed successfully PARA ERROR One of the parameters is invalid COMMENTS The Data Frame structure must be filled out accordingly before passed to this function The image data is stored in the frame buffer one line adjacent to another without gaps The length of a line in bytes depends on the PixelFormat 8 bit Black White format takes 1 byte per pixel 16 bit YCbCr422 format takes 2 bytes per pixel The total length of a line is the number of pixels per line dx times the number of bytes per pixel If the user wants to compress an image stored in RGB color format the image data must be converted to YCbCr422 format by using the MTRGB24BitToY CbCr422 function provided in the Miscllanous Utility Group before passing it to this function If only part of the image stored in the Data Frame is to be compressed use the region to define the ROI The UpDownFlip flag is used to control if
65. or Format mat IV4 Save As Is B W Cb Cr 0x80 8 bit Black White IVA Save As Is Color 16 bit YCbCr422 16 bit YCbCr422 IV4 Save As BW B W Cb Cr 0x80 8 bit B W or Color 16 bit YCbCr422 IV4 Save As VUN B W Cb Cr 0x80 16 bit YCbCr422 or Color 16 bit YCbCr422 CompressMethod Specifies the compression method which can be one of the following pre defined macros IV4_JPEG_LOSSY Uses the JPEG Baseline Compression IV4 JPEG LOSSLESS Uses the JPEG Lossless Compression QualityFactor Specifies a Quality Factor which ranges from 1 to 255 when IV4_JPEG_LOSSY is used When IV4_JPEG_LOSSLESS is used it selects one of the predictors to be used in compression The valid value is from 1 to 7 Comments Pointer to a byte buffer that contains application related information such as comment time stamp etc The contents in the buffer will be copied to the compressed file using a JPEG Comment Marker The user can retrieve this information by calling either the MTInquireJPG Vals or the IV4JPEGLoad cLength Specifies the length of the information in bytes If the contents is longer than cLength the extra part will be dropped The user must ensure that the cLength is not longer than the size of the Comments buffer RETURN VALUE IV4 OK The function completed successfully IVA ERROR The board has not been initialized JPEG Software Development kit User Guide Version 2 0 3 36 Function References IV4_NOT
66. out introducing any dis tortion The decompressed image will be exactly the same as the orig inal JPEG Software Development kit User s Guide Version 2 0 1 2 Introduction 1 2 Baseline Compression and Decompres sion The Baseline compression decompression is based on Discrete Cosine Trans form DCT Quantization and Huffman coding Block diagrams of these pro cesses are shown in Figure 1 1 and Figure 1 2 Compression To achieve the best compression results JPEG handles colors as separate com ponents such as RGB and YCbCr YUV For Baseline MuTech JPEG always converts RGB images into YCbCr color space for compression The decom pressed image can be converted back from YCbCr to RGB if so desired by the application After the pre process of color space conversion if necessary the image will go through the following procedures as shown in Figure 1 1 for compression 1 Animage is divided up into 8x8 blocks Each block is transformed into the frequency domain using Forward DCT The results of the transfor mation is a set of 64 DCT coefficients One of these values at location 0 0 1s referred to as the DC coefficient and the rest as the AC coeffi cients 2 Each of these DCT coefficients is quantized using one of the corre sponding values from the quantization table This quantization table can be scaled to generate different amount of compression The scaling process is controlled by the Quality Factor provided by t
67. ption to disregard the comment field JPEG Memory2Memory Compress JPEG File2Memory Compress See the example for JPEG_Memory2Memory_Compress 3 3 7 MTInquireBMP Vals SYNOPSIS DESC RIPTIO N PARAM ETERS RETURN VALUE Version 2 0 include mtjpeg h int MTPROCALL MTInquireBMPVals BYTE FAR file UWORD FAR dx UWORD FAR dy UWORD FAR fmt Inquires the image size and format in a specified BMP file file Specifies the file name The file name can contain a path and must have an extension of BMP dx dy Pointer to UWORD variables that return the size of the image in terms of pixels and lines in the BMP file fmt Pointer to an UWORD variable that returns the color depth of the image in the BMP file The possible values returned are 8 or 24 bits OK The function completed successfully PARAM ERR file does not have an extension of BMP JPEG Software Development Kit User Guide COMMENTS SEE ALSO EXAM PLE Function References 3 47 The application can call this function to get the image size and the data format about the BMP file It is helpful that the application need to knows the information before further operates on that file JPEG File2Memory Compress See the example for JPEG_File2Memory_Compress 3 3 8 MTInquireTGAVals SYNOPSIS DESC RIPTIO N PARAM ETERS RETURN VALUE COMMENTS SEE ALSO EXAMPLE include mtjpeg h int MTPROCALL MTIn
68. quireTGAVals BYTE FAR file UWORD FAR dx UWORD FAR dy UWORD FAR fmt Inquires the image size and format in a specified TGA file file Specifies the file name The file name can contain a path and must have an extension of TGA dx dy Pointer to UWORD variables that return the size of the image in terms of pixels and lines in the TGA file o fmt Pointer to an UWORD variable that returns the color depth of the image in the TGA file The possible values returned are 8 or 24 bits OK The function completed successfully PARAM ERR file does not have an extension name of TGA The application can call this function to get the image size and the data format about the TGA file It is helpful that the application knows the information before further operates on that file JPEG File2Memory Compress See the example for JPEG_File2Memory_Compress JPEG Software Development kit User Guide Version 2 0 3 48 Function References 3 3 9 MTInquireTIFF Vals SYNOPSIS include mtjpeg h int MTPROCALL MTInquireTIFFVals BYTE FAR file UWORD FAR dx UWORD FAR dy UWORD FAR fmt DESCRIPTION Inquires the image size and format in a specified TIF file PARAMETERS file Specifies the file name The file name can contain a path and must have an extension of TIF dx dy Pointer to UWORD variables that return the size of the image in terms of pixels and lines in the TIF file fmt Pointe
69. r cases are not supported If the user wants to have a color image in RGB format the image data must be converted by calling the MTY CbCr422ToRGB24Bit function provided in the Miscellanous Utility Group In General there two ways that the application can retrieve the comment field from a compressed code stream The first is to call MTInquireJPG Vals function to find out the information about the comment field then allocate the buffer of the right size and retrieve the comment field during the decompression function call The second way is to blindly allocate a buffer which is large enough to hold the maximum anticipated size of the comment field then calls this function to get the information needed None JPEG Memory2File Compress MTInquireJPGVals MTRGB24BitYCbCr422 MTYCbCr422ToRGB24Bit See sample for JPEG_Memory2File_Compress 3 1 9 JPEG_Memory2Memory_Compress SYNOPSIS DESC RIPTIO N include mtjpeg h int MTPROCALL JPEG_Memory2Memory_Compress DataFrame dataFrame ROI region CompressParameter cmpPara BYTE HUGE jpegBuffer ULONG FAR bufferLength Compresses an image in a pre defined Data Frame in system memory into a memory buffer in JPG file format JPEG Software Development kit User Guide Version 2 0 3 22 Function References PARAMETERS Version 2 0 dataFrame Specifies an image in system memory as below typedef struct DATAFRAME BYTE HU
70. r found the return value will be 0 Otherwise this function will copy the comment field into the buffer pointed by Comments and return the actual length of the comment field in the variable pointed to RETURN VALUE IV4 OK The function completed successfully IVA ERROR The board has not been initialized IV4 PARAM ERR name does not have an extension name of JPG COMMENTS The application can call this function to get the image size and the data format about the JPG file and then set the Video Frame accordingly before calling to IV4JPEGLoad In general there two ways that the application can retrieve the comment field from a compressed code stream The first is to call this function to find out the information about the comment field then allocate the buffer of the right size and retrieve the comment field during the IV4JPEGLoad function call The second way is to blindly allocate a buffer which is large enough to hold the maximum anticipated size of the comment field then calls this function to get the information needed When one of the decompression function is called using the NULL pointer option to disregard the comment field SEE ALSO IV4SetVideoFrame IVAJPEGLoad EXAMPLE See the example for IVAJPEGLoad Version 2 0 JPEG Software Development Kit User Guide Function References 3 39 3 3 Misc lla neous Utility Group Functions in this group provide a variety of color space conversion or file inform
71. r to an UWORD variable that returns the color depth of the image in the TIF file The possible values returned are 8 16 or 24 bits RETURN VALUE OK The function completed successfully PARAM ERR file does not have an extension of TIF COMMENTS The application can call this function to get the image size and the data format about the TIF file It is helpful that the application knows the information before further operates on that file SEE ALSO JPEG File2Memory Compress EXAMPLE See the example for JPEG File2zMemory Compress Version 2 0 JPEG Software Development Kit User Guide Glossary A 1 AppendixA Glossary compression compression data DCT coefficient decode dequantization disc rete cosinetransform DCT Huffman decoding Huffman encoding Huffman table J PEG lossless lossy JPEG Software Development kit User s Guide Reduction in the number of bits used to represent source image data Either compressed image data or table specification or both The amplitude of the specific cosine basis function A process which takes the input compressed image data and outputs a contin uous tone image The inverse procedure to quantization by which the decoding process recovers a representation of the DCT coefficients Either the forward discrete cosine transform or the inverse discrete cosine transform An entropy decoding procedure which recovers the symbol from each variable length co
72. rdization Schedule B 9 C Prediction Version 2 0 JPEG Software Development Kit User s Guide Introduction 1 1 1 Introduction 1 1 Overview MuTech s JPEG image compression Software Development Kit SDK is based on an international standard format the Joint Photographic Experts Group JPEG or ISO DIS 10918 1 The SDK provides a highly optimized DCT algorithm and high speed software that compresses gray scale or color images to ratios as high as 100 1 OEMs and System integrators can quickly andeasily incorporate MuTech JPEG SDK functions into their applications for Windows or DOS Some of these functions work independently and some of them work in conjunction with MuTech s frame grabber products and corre sponding SDKs The JPEG SDK includes libraries DLLs sample source code and build make files The key features of the MuTech JPEG SDK are m High level functions with sample code to speed up application devel opment Selectable input output file formats Quality factors 1 Qf 255 Lossy DCT based or Lossless compression modes The code stream generated can be stored either in file or in memory buffer MuTech JPEG SDK provides a choice of lossy or lossless compression modes W The Lossy also known as Baseline compression allows a selectable amount of loss of information during compression by adjusting the Quality Factor QF parameter W The lossless mode compresses the image with
73. reel J PEG E Software Development Kit User s Guide MuTech Corporation 85 Rangeway Road Billerica MA 01862 USA For additional assistance call MuTech s Technical Support Group at Telephone 978 663 2400 Fax 978 663 3444 Internet support mutech com Website www mutech com Image VGA 400 IV 400 M Vision 500 MV 500 M Vision 1000 MV 1000 are trademarks of MuTech Corporation Copyright 1999 MuTech Corporation All rights reserved Revision 2 0 4 30 99 Table of Contents 1 Table of Contents 1 Introduction 1 1 OVEIVIEW a uxo PEE LAE PAAR A DE wee eee a eee EE 1 1 1 2 Baseline Compression and Decompression 1 2 1 3 Lossless Compression and Decompression sess 1 3 2 Using MuTech PEG SDK 2 1 DOS T6 Bit Library 4 a xu wear x E eC ete em n TO Pee dnce pe 2 1 22 Windows 16 Bit DLL s 2 2 2 3 Windows32Bit DLL 2s 2 3 3 Function References 3 1 Application Level Group eer 3 GM sare oso UR e eR SS ee eno 3 1 3 11 JPEG InquireSDKVersion 0 0 0 e 3 2 3 1 2 JPEG InquireCodeStreamSize 0 eee eee 3 3 3 1 3 JPEG Hile2File Compress 0 0 0 0 ccc cece ence eens 3 4 3 1 4 JPEG Hile2File Decompress 0 0 0 0 cece III 3 6 3 1 5 JPEG File2Memory Compress sss 3 8 3 1 6 JPEG Memory2Hile Decompress 3 11 3 1 JPEG Memory2Hile Comptes 3 13 3 1 8 JPEG File2Memory Decompress seeeeee e 3 18 3 1 9 JPE
74. s must use the same set Huffman tables may be pre defined and used by an application as defaults or computed specifically for the image in an initial statistics gathering pass Such choices are the business of specific applications and outside the domain of the JPEG standard which spec ifies no default Huffman or quantization tables In contrast Arithmetic coding 5 requires no tables to be externally input be cause it adapts to the image statistics as it encodes the image in a single pass Furthermore Arithmetic coding has produced 5 10 better compression than Huffman for many of the images which the JPEG committee has tested How ever it is somewhat more complex than Huffman coding and may be less fea sible than Huffman for the highest speed hardware implementations For color images with moderately complex scenes all DCT based modes of operation typically produce the following levels of picture quality for the indi cated ranges of compression These levels are only a guideline quality and compression can vary dramatically according to source image characteristics and scene content W 0 25 0 5 bits pixel moderate to good quality sufficient for some ap plications m 0 5 0 75 bits pixel good to very good quality sufficient for many ap plications 0 75 1 5 bits pixel excellent quality sufficient for most applications 1 5 2 0 bits pixel usually indistinguishable from the originals suffi cient for the most deman
75. se will result in a final International Standard which the committee hopes to achieve before the end of 1991 JPEG Software Development kit User s Guide Version 2 0 B 10 Overview of the JPEG Still Picture Compression Algorithm Version 2 0 JPEG Software Development kit User s Guide Prediction C 1 Appendix C Prediction Figure C 1 shows the relationship between the positions a b c of the recon structed neighboring samples used for prediction and the position of x the sample being coded Figure 3 1 Relationship between Sample amp Prediction Samples Define Px to be the prediction and Ra Rb and Rc to be the reconstructed sam ples immediately to the left immediately above and diagonally to the left of the current sample The allowed predictors one of which is selected in the scan header are listed in Table C 1 JPEG Software Development kit User s Guide Version 2 0 C 2 Prediction PREDICTOR NUMBER PREDICTION 0 No prediction 1 Px Ra 2 Px Rb 3 Px Rc 4 Px Ra Rb Re 5 Px Ra Rb Rx 2 6 Px Rb Ra Rc 2 7 Px Ra Rb 2 Table 0 1 Predictors for Lossless Coding Selections 1 2 and 3 are one dimensional predictors and selections 4 5 6 and 7 are two dimenstional predictors The one dimensional horizontal predictor prediction sample Ra is used for the first line of samples at the start of the scan and at the beginning of each re
76. sed code stream A NULL pointer could be used if the user has no interest of the comment content at the time of calling If a buffer pointer is specified its length must be specified by cLength see below for detailed description It is the caller s responsibility to JPEG Software Development kit User Guide Version 2 0 3 12 Function References allocate enough space for the buffer otherwise the extra part of the Comments field will be discarded The user can also retrieve the comment information by MTInquireJPGInMemoryVals function cLength Pointer to an UWORD variable that can be used in a number of ways described below W points to NULL the user is not interested in any information in the comment field The function will not look for Comment Marker and totally ignore the Comments buffer E points to a variable with the value 0 The user is interested in only the length of the comment field If there is no Comment Marker found the return value will be 0 unchanged Otherwise this function will return the actual length of the comment field in the variable pointed to The Comments buffer will be ignored E points to a variable with non zero value The user is interested in both the length and the content of the comment field In this case the Comments buffer must be allocated and supplied and the value of the variable must be equal to the size of the Comments buffer If there is no Comment Marker found the return va
77. ser has no interest of the comment content at the time of calling If a buffer pointer is specified its length must be specified by cLength see below for detailed description It is the caller s responsibility to allocate enough space for the buffer otherwise the extra part of the Comments field will be discarded The user can also retrieve the comment information by calling MTInquireJPG Vals function cLength Pointer to an UWORD variable that can be used in a number of ways described below W points to NULL the user is not interested in any information in the comment field The function will not look for Comment Marker and totally ignore the Comments buffer E points to a variable with the value 0 The user is interested in only the length of the comment field If there is no Comment Marker found the return value will be 0 unchanged Otherwise this function will return the actual length of the comment field in the variable pointed to The Comments buffer will be ignored E points to a variable with non zero value The user is interested in both the length and the content of the comment field In this case the Comments buffer must be allocated and supplied and the value of the variable must be equal to the size of the Comments buffer If there is no Comment Marker found the return value will be 0 Otherwise this function will copy the comment field into the buffer pointed by Comments and return the actual length of the
78. start interval The selected predictor is used for all other lines The sample from the line above prediction sample Rb is used at the start of each line except for the first line At the beginning of the first line and at the beginning of each restart interval the prediction value of 2P 1 is used where P is the input precision Version 2 0 JPEG Software Development Kit User s Guide Index A AC coefficient llle lee 1 2 B 4 Arithmetic coding B 5 B baseline 1 1 1 3 3 5 3 16 3 35 B 2 B 7 build up PLOSTESSIVE usse eerte an B 7 Sequential eiie oie ERU B 7 C cod stirean EE 1 1 Code stream siZe ce CERE S 3 3 color space conversion o 1 2 3 39 Comment Marker 3 5 3 7 3 12 3 15 3 20 3 24 3 29 3 32 3 35 3 38 3 43 3 45 compressed code stream 1 4 3 9 3 12 Compressed Data Format 3 5 3 7 3 15 3 19 3 21 3 28 compression tiethiod icc ern ttt etti di 3 4 DROE eegne e E 3 4 compression mode A iet RE 1 1 1 5 A 1 B 3 aci n 1 1 A 1 compression ratio eee eee 3 36 D Data Frame 3 133 15 3 18 3 19 3 22 3 27 DC coefficient ooooooooooooooo o 1 2 B 4 DET Ad 1 1 1 2 A 1 B 3 DCT coeficent llle Al Decompressed Data Format 3 7 3 19 dequantization 0 cece eee eee es Al AS EEN 3 2 Differential Pulse Code Modulation 1 3 Discrete Cosine Transform ooooo ooo 1 2 BI H PEE ette
79. start value Usually this value will generate acceptable results for both compression ratio and compressed image quality The user may experiment with different settings and choose the best one For lossless compression a Predictor Number is used For details about the definition of the Predictor Number refer to Appendix C This function supports three type of input files which are BMP TGA and TIF Moreover only 8 bit Black White and 24 bit RGB image can be compressed except for TIFF file in which case the YCbCr422 format is also supported The output file must be JPG file Compression Data Format will be determined internally by the source data file and the Compress Method If the input data is 8 bit Black White the Compressed Data Format will be 8 bit Black White If the input Data is 24 bit RGB for lossless compression the Compressed Data format will be 24 bit RGB for baseline compression the input data will be converted to YCbCr422 and the Compressed Data Format will be YCbCr422 For YCbCr422 input data format only lossy compression is supported and it will be compressed in the YCbCr422 format SIDE EFFECT None SEE ALSO JPEG File2File Decompress MTInquireJPGVals JPEG Software Development Kit User Guide Version 2 0 3 6 Function References EXAM PLE CompressParameter cPara fill out the input parameter structure cPara CompressMethod JPEG LOSSY cPara QulityFactor 128 cPara CompressedDataFormat
80. t En abico et 2 2 JPEG Software Development kit User s Guide Index 1 Windows Ion 2 2 Windows 32 Bu 2 3 DPCM EE 1 3 Dynamic Link Library 2 2 E Entropy coding 00 00 0002 eee B 5 F Forward DET EELER as 1 2 B 4 Function INAlnourel DON ale 3 37 IVAJPEGbLoad eene 3 30 IVAJPEGSavVe E 3 34 JPEG Hile2File Compress sees 34 JPEG Hile2File Decompress 3 6 JPEG File2Memory Compress 3 8 JPEG File2Memory Decompress 3 18 JPEG_InquireCodeStreamSize 3 3 JPEG_InquireSDKVersi0M s 3 2 JPEG_Memory2File_Compress 3 13 JPEG_Memory2File_Decompress 3 11 JPEG Memory2Memory Compress 321 JPEG_Memory2Memory_Decompress 3 27 MfnourebM DV als 3 46 MTInquireJPGInMemory Vals 3 44 MTInquireJPGVals eene 3 43 MTInquireTGAV ls eee 3 47 MTInquireTIFFV als eee 3 48 MTRGB2ABitToYCbCr422 sseessss 3 39 MTRGB32BitToYCbCr422 esessss 3 41 MTYCbCr422RGB32Bit esses 3 42 MTYCbCr422ToRGB2ABit ssss 3 40 G Group Application Level esee 3 1 Version 2 0 2 Index Developer Level eere es 3 1 Miscellaneous Utility esses 3 1 Miscellanous Utility
81. terleaved in order to limited to four each the number of Q tables and sets of Huffman tables or if Arith metic coding is employed the number of statistics areas which must be stored in the decoder Baseline System 4 is the name given to a restricted version of the sequential build u mode of DCT compression It is required to be present in all DCT based modes of operation in order to provide a fall back or default mode for applications like facsimile in which sender and receiver can negoti ate a set of common features prior to image encoding and transmission But the Baseline System is also expected to be sufficient for many applications in its own right Its restrictions with respect to general sequential operation are a Operates on images with 8 bits pixels component only b Uses Huffman coding only c Uses at most two sets of Huffman tables per scan These restrictions are aimed at reducing the cost of the codec a reduces the computation cost of the FDCT and IDCT as described in section 3 b reduces the options available for entropy coding and c reduces the storage cost re quired at the decoder B 6 Progressive Build up The mode of operation known as progressive build up 6 is in many ways identical to sequential build up The difference is that each image component is encoded as multiple scans rather than as a single scan In this mode a rough but recognizable version of the image appears quickly in co
82. tors to be used in compression The valid value is from 1 to 7 B CompressedDataFormat This entry is reseved and must be filled with the following pre defined macro JPEG AS SOURCE See comments for details JPEG Software Development Kit User Guide Function References 3 5 E Comments Pointer to a byte buffer that contains application related information such as comment time stamp etc The contents in the buffer will be copied to the compressed file using a JPEG Comment Marker The user can retrieve this information by calling either the MTInquireJPG Vals or one of the decompression functions that decompresses from a JPG file E CommentLength Specifies the length of the information in bytes If the contents is longer than CommentLength the extra part will be dropped The user must ensure that the CommentLength is not longer than the size of the Comments buffer OutFile A pointer to a string that specifies the output file name which may contain a path and must have an extension of JPG RETURN VALUE OK The function completed successfully PARA ERROR One of the parameter is invalid COMMENTS For lossy baseline compression the Quality Factor is used to control the compressed image quality The higher the Quality Factor the better the compressed image quality and the lower the compression ratio and vice versa To adjust the compression ratio and compressed image quality MuTech recommends using 128 as the
83. umber of lines in the Data Frame B UpDownFlip This flag specifies the direction of compressing inverse vertically if set to 1 See the example code for MTRGB2ABitY CbCr422 function on how to use this flag B PixelFormat Specifies the pixel format for the Data Frame Only the following formats are supported DATA 8bit BW 8 bit per pixel Black White format DATA l16bit YCbCr422 16 bit per pixel YCbCr422 format DATA l6bit YONLY 16 bit per pixel YCbCr422 format But only the Y component is relavent in this case the Cb and Cr component will be filled with values of 0x80 DATA AS COMPRESSED Specifies that the Decompressed Data Format format is determined by the Compressed Data Format in the input file When the Compressed Data Format is 8 bit Black White the Decompressed Data Format will be DATA 8bit BW When the Compressed Data Format is 16 bit YCbCr422 the Decompressed Data Format will be DATA 16bit Y CbCr region Specifies a Region Of Interest by the ROI structure typedef struct ROI UWORD sx UWORD sy UWORD dx UWORD dy ROT The members of the structure are explained below W sx sy Specify in term of pixels the offset of the upper left corner of the region in a Data Frame B dx dy Specify in term of pixels and lines the width and height of the region in a Data Frame Version 2 0 JPEG Software Development kit User Guide Function References 3 29 Note a region must be assoicated
84. variable pointed to The Comments buffer will be ignored E points to a variable with non zero value The user is interested in both the length and the content of the comment field In this case the Comments buffer must be allocated and supplied and the value of the variable must be equal to the size of the Comments buffer If there is no Comment Marker found the return value will be 0 Otherwise this function will copy the comment field into the buffer pointed by Comments and return the actual length of the comment field in the variable pointed to RETURN VALUE OK The function completes successfully PARA ERROR One of the parameter is invalid COMMENTS This function is functionally identical to JPEG_File2Memory_Decompress The only difference is that this function reads the compressed data from a system buffer instead of a disk file The files generated by these two functions are identical Please refer to JPEG_File2Memory_Deompress for comments JPEG Software Development Kit User Guide Version 2 0 3 30 Function References SIDE EFFECT SEE ALSO EXAM PLE In General there two ways that the application can retrieve the comment field from a compressed code stream The first is to call MTInquireJPGInMemory Vals function to find out the information about the comment field then allocate the buffer of the right size and retrieve the comment field during the decompression function call The second way is to blindly
85. w either 8 bit or 12 bit per component input pixel samples knowing that the high er precision will require higher cost computation but will accommodate addi tional applications After quantization the DC coefficient is treated separately from the 63 AC co efficients Because there is typically strong correlation between the DC com JPEG Software Development kit User s Guide Overview of the JPEG Still Picture Compression Algorithm B 5 ponents of adjacent 8x8 blocks the quantized DC term is represented by encoding its difference from the DC term of the previous block The special treatment is worthwhile as the DC terms typically contain a significant frac tion of the total image energy Entropy coding is the final processing step Whereas compression is achieved in the quantization step by discarding data which is not visually significant en tropy coding achieves additional compression without additional loss by en coding the data more compactly based on its statistical characteristics Except for the Baseline System in which only Huffman coding is allowed either Huffman coding or Arithmetic coding may be used as the entropy coding method for all DCT based modes of operation Huffman coding 4 requires use of one or more sets of Huffman code tables A set consists of one table for the DC coefficients and one table for the AC co efficients Each image component may use only one set of Huffman tables though not all component
86. with values of 0x80 DATA AS COMPRESSED Specifies that the Decompressed Data Format format is determined by the Compressed Data Format in the input file When the Compressed Data Format is 8 bit Black White the Decompressed Data Format will be DATA 8bit BW When the Compressed Data Format is 16 bit YCbCr422 the Decompressed Data Format will be DATA 16bit YCbCr422 region Specifies a Region Of Interest by the ROI structure typedef struct ROI UWORD sx UWORD sy UWORD dx UWORD dy ROI The members of the structure are explained below E sx sy Specify in term of pixels the offset of the upper left corner of the region in a Data Frame B dx dy JPEG Software Development Kit User Guide Version 2 0 3 20 Function References RETURN VALUE COMMENTS Version 2 0 Specify in term of pixels and lines the width and height of the region in a Data Frame Note a region must be assoicated with a Data Frame in order to be completely defined The region is used to put the decompressed image into an ROI of a larger Data Frame To decompress the image into entire Data Frame define the region equal to the frame If the ROI is smaller than the image size only the upper left part which is equal to region will be filled into the Data Frame lt Comments Pointer to a byte buffer that will be used to hold the JPEG comment information if one exists in the compressed code stream A NULL pointer could be used if the u
87. xists in the compressed code stream A NULL pointer could be used if the user has no interest of the comment content at the time of calling If a buffer pointer is specified its length must be specified by cLength see below for detailed description It is the caller s responsibility to allocate enough space for the buffer otherwise the extra part of the Comments field will be discarded The user can also retrieve the comment information by MTInquireJPGVals function cLength Pointer to an UWORD variable that can be used in a number of ways described below W points to NULL the user is not interested in any information in the comment field The function will not look for Comment Marker and totally ignore the Comments buffer E points to a variable with the value 0 The user is interested in only the length of the comment field If there is no Comment Marker found the return value will be 0 unchanged Otherwise this function will return the actual length of the comment field in the variable pointed to The Comments buffer will be ignored E points to a variable with non zero value The user is interested in both the length and the content of the comment field In this case the Comments buffer must be allocated and supplied and the value of the variable must be equal to the size of the Comments buffer If there is no Comment Marker found the return value will be 0 Otherwise this function will copy the comment field into the
88. zero value The user is interested in both the length and the content of the comment field In this case the Comments buffer must be allocated and supplied and the value of the variable must be equal to the size of the Comments buffer If there is no Comment Marker found the return value will be 0 Otherwise this function will copy the comment field into the buffer pointed by Comments and return the actual length of the comment field in the variable pointed to RETURN VALUE OK The function completed successfully JPEG Software Development Kit User Guide Version 2 0 3 46 Function References COMMENTS SEE ALSO EXAM PLE PARAM ERR jpegBuffer is NULL The application calls this function to get the image size and the data format in a memory buffer in JPG file format It is helpful that the application knows the information before further operates on the memory buffer In General there two ways that the application can retrieve the comment field from a compressed code stream The first is to call this function to find out the information about the comment field then allocate the buffer of the right size and retrieve the comment field during the decompression function call The second way is to blindly allocate a buffer which is large enough to hold the maximum anticipated size of the comment field then calls this function to get the information needed When one of the decompression function is called using the NULL pointer o

Software Development Kit User's Guide

Contents

Download Pdf Manuals

Related Search

Related Contents