User Manual - booksorber.com
Contents
1. Detect Boundary Subdivide Figure 4 3 Button Bar with most important commands The progress bar indicates the position of the currently selected scan within the book Figure 4 4 Typical Bezier curve with control Points for a book page boundary s e Repeat these steps for the next 2 3 scans navigate there using the navigation buttons in the button bar and also for the last few scans of the book 18 19 4 2 Getting Started 15 e Press the Subdivide button or press the D hotkey for divide to jump to the middle of the book and again set the boundary Repeat this procedure until all boundaries are sufficiently accurate Successive Subdivide commands will automatically jump to positions refining the intervals between scans with explicitly set boundaries giving the best result with minimal work Tip You can use the demo project included with the distribution automatically opens up when the program is opened for the first time for the first experiments without the need to provide your own scans 4 2 3 Making Corrections Unwanted scans blurred image hand in image can be removed through Edit gt Delete Scan 4 2 4 Generating the final output In the Edit gt Global Settings dialog the number of program threads used for parallel processing of the images can be modified It has to equal at least 1 and should be no more than the number of available processor cores The choice de termines a comprom2. booksorber User Manual Booksorber User Manual v1 0 3 Typesetting The Author Day of Print April 22 2013 Booksorber in a nutshell Booksorber is an innovative new software which allows you to quickly and easily digitize your books for mobile use employing a digital camera No further invest ment in expensive hardware is needed and it only takes a few minutes to process an entire book in excellent quality Contents 1 1 1 2 1 3 2 1 2 2 2 3 3 1 3 2 3 3 4 1 4 1 1 4 1 2 4 1 3 4 1 4 4 1 5 4 1 6 4 2 4 2 1 4 2 2 4 2 3 4 2 4 4 2 5 5 1 5 2 6 1 6 2 6 3 Introduction Disclaimer ui as 1 Thanks for using Booksorber ooooocoocccocooco o 1 Hardware Requirements oooococcccoco o 1 Software Installation a A A II ee a ae 3 Mac OSX susanne anne ae re 3 ERE Er 4 Capturing the raw image material Scanning Rig Setup 1 eee ee 5 Camera Settings tetera tcs 6 Efficient Scanning ooooooccocoo tte ee 8 Processing with Booksorber Ihe Concepts rar a Y A eR eshte ee das 11 Dewatping csi ida speed nama iii 11 Lighting correction ms rosarina 11 Text and Image Quality Enhancement 12 Finger Removal 224c02i ied rro oca 12 PDF BIE eozacuorss cara bas re sia ae a 12 PDF Post Processing 2 4 2u 20H A en bab awe wh aed 12 Getting Started emitan te ont Semaine eae ess 12 Creating a new Project 0 0 cece eee eee eee 12 Project Set p simane ees ou e
3. Acknowledgements Many thanks to the early on supporters of this product Mag Thomas Geif Granadia Paul Martschini Martin Kubat Andre Schopper Unger Ernst Evan Robinson Daniel Blauensteiner Adrien Welter Anthony Moscar Mike Buebe Florian Tr xler Andreas Job Akos Kaszoni Gerd Dickmeiss Hinrich Peters Gal Buki Mark Klingman Alex Lane The program distributed with this documentation includes open source third party libraries governed by licences included within the licences directory of the distri bution 23 Contact and further Info Booksorber Homepage http www booksorber com email office booksorber com 25
4. Linux which can make use of all available memory e When using the Windows Booksorber exe executable a proper amount of memory is automatically allocated in this case e OSX and Linux users have to edit the respective launch files with a standard text editor and manually change the usable amount of memory to a suitable value e g set Xmx3000m to use 3000 MB A good choice would be to use your available memory minus 1000MB to leave some room for other system processes Never allocate more than physically available memory 5 2 Slow Processing Speed Depending on the size of the raw images and the number of available processor cores processing currently takes something around 10 seconds per page on average Please don t be afraid if processing of a few pages seems to take much longer some lengthy one time pre calculations are performed at startup can last up to 2 minutes no output is generated during that phase Processed pages are written to the out folder of the project directory on the fly Processing at least 8 pages and comparing the timestamps of the first and last processed page gives a more precise estimate for the average processing time on your system 19 20 5 Troubleshooting The performance will be further improved in future releases Nevertheless Book sorber should already be quite useful since it works unattended once it s set up Capturing your book using a digital camera is really fast and already now you c
5. processing threads at the expense of processing speed see the Troubleshooting Section 5 1 for details Launch by executing Booksorber exe 2 2 Mac OSX Starting with OSX 10 7 the Java Runtime is no longer preinstalled and has to be manually installed before using Booksorber See this link http helpx adobe com x productkb global install java jre mac os html for details Please make sure you use a 64bit kernel and Java runtime if possible If you en counter OutOfMemory Errors in that case please edit the launch script and increase the amount of memory Booksorber is allowed to use see the Troubleshooting Sec tion 5 1 for details On a 32bit system you can at least circumvent the error by reducing the number of processing threads Edit gt Global Options at the expense of processing speed Launch by executing launch macosx command 4 2 Software Installation 2 3 Linux Please make sure a current Java Runtime JRE 6 or higher is installed Please make sure you use a 64bit kernel and Java runtime if possible If you en counter OutOfMemory Errors in that case please edit the launch script and increase the amount of memory Booksorber is allowed to use see the Troubleshooting Sec tion 5 1 for details On a 32bit system you can at least circumvent the error by reducing the number of processing threads Edit gt Global Options at the expense of processing speed Launch by executing launch linux sh 3 Capturing the raw i
6. 09 03 28 17 44 16 288x2 highlights are clipped 3 3 Efficient Scanning e The basic idea is simple Take an image turn the page take an image turn the page A speed of 80 pages per minute can be achieved It s most convenient to look at the book from the correct side like you would read it This means that with the described setup images are taken upside down which is no problem because they can later be rotated by 180 in the Booksorber software e Pull the pages slightly apart with two fingers in the middle of the left and right border as shown in Figure 3 2 Although corrected later less curvature near the spine means higher quality results Warning Never put fingers on the top border bottom border or near the cor ners 18 e Hold the trigger in one hand permanently and turn the pages with the other one as illustraded in Fig 3 4 Also see the video on the homepage http www booksorber com index php page media video Tip If using an IR trigger without direct line of sight to the camera sensor triggering is more reliable by directing it towards the book which then reflects the IR light 13 This interferes with automatic boundary detection 3 3 Efficient Scanning 9 e Take care to turn the pages consistently without abrupt moves between taking pictures Figure 3 4 Example of typical handling 14 15 4 Processing with Booksorber 4 1 The Concepts 4 1 1 Dewarp
7. PDF file which is then automatically opened in the default viewing application The PDF export step can be repeated afterwards e g to choose different cropping borders without the need to reprocess all pages by selecting Process gt Re Export PDF 4 2 5 Help This manual can be displayed from within the program Help gt Help Contents 4 2 Getting Started 17 Figure 4 6 Pdf Export Dialog A transparent overlay of all processed pages is shown to make selections of cropping Borders easy 5 Troubleshooting 5 1 Out of Memory Error Parallel Processing of high resolution images is a memory intensive task Book sorber memory requirements depend linearly on the number of pixels of the raw images and the chosen number of processing threads determining the number of images processed in parallel If using the 32bit release of Booksorber Windows or an external 32bit Java Run time Environment OSX Linux only 1 2 1 5 GB of RAM can be used due to constraints of 32bit Java programs so it is likely to get an OutOfMemory Error when processing larger images with the default settings It is always possible to get rid of this error by reducing the number of processing threads under Edit gt Global Settings at the expense of lower processing speed If on a 64bit operating system the best solution to the problem would be to also use the 64bit release of Booksorber Windows or an external 64bit Java Runtime Environment OSX
8. Probably the best but also expensive tool for OCR and PDF post processing is Acrobat Professional by Adobe the original developers and maintainers of the PDF standard Using their newest ClearScan technology text is not only recognized but replaced by a vectorized font generated on the fly from the averaged image input which yields smooth print quality at any zoom level and much smaller files Tip PDF XChange Viewer from Tracker Software http www tracker software com product pdf xchange viewer also performs OCR reliably and is free for pri vate and commercial use Other well known commercial products are e Abby FineReader http finereader abbyy com e Nuance OmniPage http www nuance com for individuals by product omnipage index htm For more see e This exhaustive list of OCR software on the english Wikipedia http en wikipedia org wiki List of optical character recognition software or 21 20 22 6 Tips for PDF Post Processing and OCR e A more concise comparison on the german Wikipedia http de wikipedia org wiki Texterkennung 0CR Software Unfortunately open source solutions are not quite mature yet OCR also is quite a complex topic 6 3 End formats other than PDF PDF is a format very well suited for digitization of complex print material like college and university textbooks with mixed text photos tables formulas and vector illustrations In particular viewers are availab
9. The leg won t throw a visible shadow if the light is large and close enough Tip When using a tripod painted in black a piece of aluminum foil wrapped around it near the light source can prevent heat damage Place the light source in a slightly elevated position e g staple it on some books so that the angle of incidence of the light is not completely flat e Reduce ambient light from other light sources close curtains move away from strong room lights so that they are negligible compared to the main light source After correctly adjusting exposure for the main light a picture taken with the main light switched off should essentially be black or at least very dark 3 2 Camera Settings To ensure consistent results in the post processing step all book pages have to be shot under the same conditions Therefore all relevant camera settings have to be done manually In the following all necessary steps are described conceptually refer to your camera s user manual if you re not sure on how they translate to your specific model Image Fileformat Select the highest quality and highest resolution JPG output option A too shallow angle of incidence means that unevenness in the paper structure is strongly accented due to shadows which is undesirable Again this is necessary to prevent shadows at the book spine due to sideway illumination Also very strong light gradients should be avoided e g sunlight
10. an finish post processing of a 300 page book in an hour while having lunch In some cases Booksorber might not automatically recognize the number of virtual cores available on your computer If you experience slow performance on a modern multi core processor please check the number of processing threads under Edit gt Global Settings It makes sense to increase this value to a maximum of the number of available virtual cores which you can determine by checking the task manager activity monitor of your system The preset value defaults to the maximum value minus one to keep your system responsive during processing 6 Tips for PDF Post Processing and OCR 6 1 The benefits of OCR Just as for any other book scanner quality enhanced images of the book pages are the primary output of Booksorber Even if they are assembled into PDF file format for convenient viewing they stay just that bitmap images Such PDF files also take up quite a bit of storage space estimated at about 1MB per page If one wishes to search within the text or easily highlight text sections it is necessary to perform OCR optical character recognition on the output using readily available third party tools Depending on the chosen tool this higher level interpretation can also visibly enhance text quality and most importantly significantly reduce the file size of the final PDF typical shrinkage by a factor of 10 6 2 Tools for OCR commercial and free
11. e ee hanes 12 Making Corrections 0 000 ccc eee ce eee eee 15 Generating the final output 00000002 eee eee eee 15 A oe se a E tees E E eon eee ee same 16 Troubleshooting Out of Memory Error teed 24 ras anne 19 Slow Processing Speed ccc cede ceeded aan cues ase baw een 19 Tips for PDF Post Processing and OCR The benefits of OCR cats ai 21 Tools for OCR commercial and free 21 End formats other than PDF odiar rea Banana 22 vi Acknowledgements Contact and further Info DH w 1 Introduction 1 1 Disclaimer The software shipped with this manual is the first official release version This means e Not all envisioned features are integrated yet E g finger recognition is still missing and will be delivered with one of the following updates e Processing speed will still be significantly increased in the future by introducing OS specific optimizations e Despite thourough testing and use of solid design principles software bugs can slip into the product In any case we try hard fix any issues upon feedback as soon as possible with a new update 1 2 Thanks for using Booksorber Thank you for choosing Booksorber We honestly hope to deliver a useful product to make high quality bookscanning affordable for everybody We are looking forward to your feedback Many thanks also to all the early on supporters and beta testers of Booksorber 1 3 Hardware Requiremen
12. e necessary for very tricky print materials like magazines with dark pictures cropped at the page boundaries 4 1 2 Lighting correction Because Booksorber intends to make book scanning possible with a minimalistic hardware setup it has to correct for uneven lighting caused by the use of a single light source To work reliably lighting conditions should nevertheless be the same for all pictures and at least 3 5 pictures of a book should be processed This works independently of book contents text graphics Interpolation works quite well because changes from one page to the next are gradual if a bit of care is taken not to suddenly shift the book while taking the pictures 11 12 4 Processing with Booksorber 4 1 3 Text and Image Quality Enhancement Text and Image sections are automatically identified and receive different treatment for optimal output quality 4 1 4 Finger Removal If it s not possible to remove fingers from the image by simple cropping in the final PDF binding stage they can be automatically detected and removed 4 1 5 PDF Binding After processing an overlay of all pages is displayed which can be conveniently used to determine crop borders before all output images are assembled to a single convenient PDF file 4 1 6 PDF Post Processing Third party tools can be easily used for text recognition or further format conver sions of the output See section 6 4 2 Getting Started Projects are
13. faces since unwanted reflections on book pages an issue with high gloss print is avoided 5 The larger surface of the light source also makes shadowless illumination of the book easier 6 The processing can be efficiently parallelized so N instead of 1 processor cores almost yield a speed increase factor of N 2 Software Installation For maximum ease of use and installation uninstallation without side effects the current versions of Booksorber are delivered as a portable application This means that installation merely consists of extracting the archive contents to some folder on your harddisk and running the main executable Nothing else on your computer program folders registry startup links will be changed The software is therefore uninstalled as easily just by deleting the extracted folder When the program is run for the first time you will be asked for your licence number please enter the one you received separately via email or choose Free Trial Mode Some special notes for different operating systems apply 2 1 Windows Please don t install Booksorber to one of the c programs or c programs x68 folders These folders have restricted access and program execution will fail because configuration files can t be written Please use the 64bit version when installing it on a 64bit operating system If only a 32bit OS is available OutOfMemory Errors may occur which can be circumvented by reducing the number of
14. ing Pages are flattened by detecting their curved boundaries and using this information to build a 3 dimensional model of the surface to correct for warping in the image The boundaries are modeled by Bezier curves with four control points known from vector graphic design programs One main feature is the automatic detection of those boundaries but they can be manually adjusted as well In the currently implemented way of operation definition of all page boundaries is done semi automatically Every few pages 10 20 the boundaries are automatically detected and manually corrected if necessary These explicitly set boundaries are indicated by red lines Boundaries of in between pages are interpolated indicated by green lines To do this in the most effective way the user is supported by a feature which pro gressively divides the intervals between those keyframes By pressing the hotkey D for divide the program successively jumps to the page in the middle the page in the middle of the first half the page in the middle of the second half and so on Manual corrections can always be done starting from the currently interpolated position of the boundaries Because of this interpolation scheme and the fact that simple Bezier curves are very well suited to describe the curvature of typical book pages it is possible to very efficiently do these steps completely manually for an entire book typically 5 minutes of work which may b
15. ise between processing speed and system responsivity during processing The number of cores minus one is used as a default which is usually fine Warning On some laptop computers extended times of 100 processor load may lead to crashes of the system because of overheating In this case the number of threads should be reduced Start the actual unattended processing by pressing Start processing Depending on your system this takes something between 10 20 seconds per page on average Tip During processing the finished output images are sequentially written to the out directory within the project directory By opening these one can monitor the output results early on and maybe terminate processing if settings need to be changed Boundaries for the first and last few pages should be set without skipping scans because at the beginning and the end of a book page positions typically change quite rapidly More possibilities e g adding scans after creation of the project and re processing only of selected scans will be included in the next releases 16 4 Processing with Booksorber ZA Project settings st ii Rotate Scans by 180 M Process in Grayscale a v Reduce image noise Detect amp Remove fingers Figure 4 5 Project Settings Menu After processing is finished the PDF export dialog is displayed Fig 4 6 Here the cropping borders for the PDF can be specified Press ENTER to export the
16. le for almost every platform including smartphones tablets and ebook readers which are mostly free and offer features like text highlighting annotation full text search bookmarks and much more Another noteworthy format which strives for something similar is the DJVU file format http en wikipedia org wiki DjVu However viewer availability and list of supported features are far inferior Besides that an entire ecosystem of ebookreader formats like EPUB exist both proprietary and open They are targeted more towards a precise high level repre sentation of mostly text only documents like novels which are then dynamically formatted depending on the size of the reading device Warning Creating ebookreader friendly output Because these formats store text as pure characters and not as some meta information overlay to an underlying scanned bitmap much work has to go into manual correction of OCR errors if one wants to generate such a file based on scanned input Tip Many tools expect one image per page as input With Booksorber these raw output files are accessible in the out folder within the project folder Especially watch out for the highlighting feature of the common WinDJView software High lighting information is stored in the program not the document file this means all highlighting information is lost upon reinstall of the software or when reading the document on another device
17. mage material 3 1 Scanning Rig Setup Please refer to section 1 3 for the required hardware components Figure 3 1 Hardware setup for scanning e Set up the tripod with the camera facing down as shown in Fig 3 1 Adjust the height of the tripod and focal length of the objective so that the book fills the frame when opened in the middle with some margin to account for movement which will occur when turning the pages during the capture process See Fig 3 2 for an example image A focal length around 50mm full format equivalent amounts to a typical distance of one meter between book and camera and is usually a good choice 7 Important to take best advantage of the available camera resolution 8 A closer distance with a more wide angle lens leads to undesirable perspective distortion 1 o o 6 3 Capturing the raw image material Tip If you have troubles getting the tripod legs out of view try extending the rear leg more but watch that the camera does not tip over Zoom objectives may have the problem of the tube creeping out focal length not staying fixed This can be prevented by fixing the focal ring by wrapping a strip of Duck Tape around it e Place the black background e Align the lightsource exactly parallel and in line with the book spine so that no shadows are produced at the spine due to sideways illumination A larger light source can be placed closely behind the rear tripod leg
18. ries of the first scan First try it automatically by pressing the Detect Boundary button or the hotkey A for automatic If manual corrections are necessary they can be done by moving the Bezier curve control points of the boundary The boundaries change from green to red as you make adjustments If automatic detection goes completely 16 A fully automatic scheme will be included in the next releases as an alternative to the semiautomatic scheme described here 17 Bezier curves function in the same manner as they do in popular image processing programs The inner control points determine the tangent of the curve and the strength of the deformation via distance from the endpoints See Fig 4 4 for an example of the typical position of the control points 14 4 Processing with Booksorber TO E osc_2320 JPGE REA E DSC_2359 JPG Ser I Wine DSC_2360 JPG Serene Map EA e DSC 2361 JPG BTe PEErA ca MOE EA IC DSC_2362 JPG BISOPELER ci Witze DSC_2363 JPG BIETE Map EA de DSC_2383 JPG DSC_2335 JPG amp DSC_2356 JPG aS ewer eye E DSC_2357 JPG OPEN TO DSC_2358 JPG Dateiname _2352 JPG DSC_2353 JPG DSC_2354 JPG DSC_2355 JPG DSC_2356 JPG Dateityp JPG Images Figure 4 2 New Project Menu wrong boundaries can be reset via Edit gt Reset Boundary or the BACKSPACE key Use the ZOOM button to toggle between the default and a magnified view first 10 previous next 10 last ZOOM
19. saved in a project folder with the following structure Main project folder containing the batch xml configuration file and the out pdf PDF output file after processing scans The raw images out Images of the processed pages 4 2 1 Creating a new Project Select Project gt New Project and specify the JPG raw images from your camera storage card in the following dialog Fig 4 2 Finally enter name and location for the project folder The images will be automatically copied there 4 2 2 Project Setup If your book is basically grayscale select the corresponding output option in the Project gt Project Settings dialog Fig 4 5 Check Uncheck 180 rotation 4 2 Getting Started 13 32 Booksorber 1001 a A wT cas Project Edit Navigate Process Help DSC_0618 JPG DSC 0653 JPG DSC_0654 JPG first 10 previous next 10 last ZOOM Detect Boundary Subdivide B Start processing K d Figure 4 1 The main program Window if necessary Automatic finger removal is activated by default There is also an option to reduce noise and JPG compression artifacts which is recommended for raw image material from cheaper cameras although it slightly increases processing times The main step is to specify the Page boundaries e All necessary commands are accessible via the button toolbar Fig 4 3 in the bottom or via hotkeys Adjust the page bounda
20. spots 3 2 Camera Settings 7 Figure 3 2 A typical image ISO Choose the lowest available ISO value sensor sensitivity typically 100 or 200 for lowest possible noise in the final image Whitebalance Choose incandescent lamp or do a manual white balance Exposure Set the camera to manual exposure Choose an aperture of 5 6 to 8 for sufficient depth of field then adjust the exposure time for proper exposure will typically end up around 1 30s Judge proper exposure by taking test shots and displaying the histogram of grey values It should look comparable to Fig 3 3 Focus Flip the book open somewhere in the middle and center align it with re spect to the viewfinder Note for DSLRs This can be done more conveniently using a Live View mode if available Use the autofocus to focus on some text section then switch to manual focus so that it won t change during the cap turing process 11 Tf using a sufficiently strong light source this should still allow reasonable exposure times The tripod also helps in this respect 12 Focus depth is of concern since the pages will always be slightly curved and the page height will vary as pages are flipped 8 3 Capturing the raw image material 1 45 NIKON D90 Sa A T mme O 1 50 F56 SO E Figure 3 3 Typical histogram of a cor AUTO 0 0 Adoben rectly exposed page The available dynamic 100NCD90 _DSC0463 NEF m pi range is well used while taking care that no 20
21. ts For efficient capturing of the raw book images e A digital camera with 10 12 Megapixels and the possibility to manually set ISO exposure white balance and focus standard features e A remote trigger cable IR wireless PC interface or your best friend Currently there are many improvements in the pipeline which will very soon be included A fixed focal length lens equivalent to a 50mm full format lens is often a good choice for reproduction work like this offering good value for money A timer function device can be useful as well to allow hands free operation Program it to take a picture approx every two seconds Functions like this can be added for some camera types using alternative firmware like the CHDK Project http chdk wikia com wiki CHDK for the Canon Powershot series 2 1 Introduction e A stable camera tripod with a head that can be tilted 90 downwards e An intense light source Strong desktop lamp may be fine but small robust halogen spotlight like the one in Fig 1 1 is recommended e A piece of matte black cloth or paper as a background For processing with Booksorber e A recent PC Laptop ideally with a dual or quadcore processor running Win dows MacOSX Linux and gt 2GB of RAM Figure 1 1 A typical small halogen spotlight used in construc tion work cheaply available at most hardware stores 4 Tf available a tripod with black finish is preferable to one with metallic sur
Download Pdf Manuals
Related Search