CIB image toolbox technical manual

1. Introduction

This library contains implementations of a number of image processing algorithms for scanned documents processing.

These algorithms can be separated into next categories:

• Filters:
• Noise filters
• Binarization filters
• Modification filters
• Affine transform filters
• Special filters
• Image information
• Mergers
• Segmenters
• Image comparers

2. Calling CIB image toolbox

Calling via CIB job: example CIB job XML

<?xml version="1.0" encoding="ISO-8859-1" ?> 

<!-- XML jobfile für CIB job from version 1.0.4 -->

<root>

<!-- Definition of Jobs or Steps that a sent to documentserver from the caller -->
<Comod>
<!-- Default settings that are valid for all Jobs and Commands, if                    
                 they are not overwritten in the job or step -->

<defaults>

 <properties command="job">
             <property name="OutputMode">XML</property>    
             <property name="UseInMemoryProcessing">1</property>
 </properties> 

</defaults> 

<!-- List of all jobs in this job --> 

<jobs>

<job name="job1" expected-result-code="0">

<!-- List of all steps that are to be executed in this Job -->

<steps>

<step name="ipl"command="image-conversion"> 

 <properties>
  <property name="InputFilename">c:/Test/1/test.tif</property> 

  <property name="OutputDirectory">c:/Test/1/</property>          
  <property name="OutputFileNamePatterns">pattern_png_###;pattern_jpeg_prev_##;
   pattern_jpeg_thumbs_##</property>
  <property name="PageSelection">{ "PageRange" : "1", "Encoding" : "PNG", };{"PageRange" : "2", "Encoding" : 

          "JPEG", };{ "PageRange" : "1-10", "Encoding" :"JPEG", "MaxWidth" : "150" }</property>          
   <property name="Overwrite">true</property>          
   <property name="Threading">{ "SelectionThreads": "3", "RenderThreads" : "3" }</property>     

</properties>
     </step>   

</steps>
</job> 

</jobs>
</Comod>

</root>

Calling via CIB runshell

cibrsh -d request.xml

3. Input CIB IPL properties

4. Output CIB IPL properties

5. Properties details

In this chapter you can find details about different properties.

5.1. InputFilename

Specifies the filename of the input file. Supported image formats: BMP, PNG, JPEG, TIFF, SVG.

Syntax

InputFilename=filename.<ext>
<ext>: Supported image formats 

 Example

InputFilename=file.tif

5.2. AlgorithmSetFileName

AlgorithmsSetFileName

String

Set

Specifies the filename of the XML defining algorithms.

Syntax

AlgorithmsSetFileName=algorithms.xml

Example

AlgorithmsSetFileName=algorithms.xml

5.3. AlgorithmSetData

AlgorithmsSetData

String

Set

Specifies the raw XML string data to be processed. This property overrides contents found on AlgorithmsSetFileName.

Syntax

AlgorithmsSetData=<XML>

Example

AlgorithmsSetData=
<algorithms>
 <algorithm name="CleanImage" inputs="InputFile=$1" outputs="CleanedImage=$0"> 
  <processor type="CleanBackgroundToWhiteFilter" inputs="InputFile" outputs="CleanedImage" />
 </algorithm>
</algorithms>

5.4. OutputDirectory

OutputDirectory

String

Set

Specifies the output directory. ‘Absolute’ and ‘relative’ specifications are allowed. The File name is specified by the property OutputFileNamePatterns.

Syntax

OutputDirectory=.\Folder\

Example

OutputDirectory=.\Folder\

5.5. OutputFileNamePatterns

With this property a pattern for output filenames can be specified.

Per pageselection there is one pattern. If no pattern is preset, a unique name is generated that also contains a number pattern for multipage files. Placeholder for numbers is the hashtag “#”, where if there are several # Symbols, zeros are filled from the left. That means: “###” leads to “000” - “099”, but the value is not limited to the number of digits as represented by # Symbols. So the value for a placeholder “###” could be “123456789”.

This value can optionally contain a relative directory (e.g. “output\manipulatedImage”). However the directory must be ‘relative’ specified.

Syntax

OutputFileNamePatterns=<pattern>[;<pattern>]
<pattern>:<Patternname>_###.<ext>
<ext>: Supported image formats

Example

OutputFileNamePatterns=pattern_png_###.tif

5.6. TraceFilename

Specifies the name of the log file (and, optionally, directory). If this property is set, a log file of the process is created. However the directory may only be defined ‘relative’.

Syntax

TraceFilename=filename.log

Example

TraceFilename=output\trace.log

5.7. PageSelection

Property for page selection. With this property you can describe, which page/pages should be processed.

If this property is not set or no parameters are passed, there is no conversion process and the logfile contains ‘Conversion isn't required: PageSelection isn't defined.’

Must only be positive values, otherwise conversion will fail with error.
‘Conversion parameter array is invalid or empty.’

Syntax

PageSelection={“PageRange”:”<start>[-<end>]”,”MaxWidth”: number, ” MaxHeight”: number, “MaxResolutionX”: number, “MaxResolutionY”:
              number, “ColorSpace”: string, “ColorDepth”: number, “Encoding”: string, [[“CompressionLevel”:number][,
              “Quality”:number] [, “Border”: { “Width”: number, “Color”: string}][, “SpecialPageRange”: 
              {“PageRange”:”<start>[-<end>]”, ”MaxWidth”: number, ”MaxHeight”: number, …}][, “ImageToolboxParams”: { “AlgorithmSetPath”: string,
              “AlgorithmName”: string,  “ProfilePath”: string]}] [, “…”:…]]};...

Example

Example #1:

PageSelection={ "PageRange" :"1", "Encoding" : "PNG", };{"PageRange" : "2", "Encoding" : "JPEG",};{ "PageRange" : "1-10", "Encoding" : "JPEG",
              "MaxWidth" : "150" }

Example #2:

{ "PageRange" :"1-3", "Encoding" : "TIFF","ColorSpace" : "sRGB", "ColorDepth" :"8", "MultiPageOutput" : true,
              "ImageToolboxParams" : { "AlgorithmSetPath" :"c:/Software/CIB/_IPL_problems/2016_11_29/AlgorithmsSet_sample.xml","AlgorithmName" : "Flip", "ProfilePath" :
              "c:/Software/CIB/_IPL_problems/2016_11_29/processing_profile.xml", },"Border" : { "Width" : "3", "Color" :"0,255,0", }, "SpecialPageRange" : { "PageRange"
              : "2-2", "AutoRotation":false, "MaxHeight" :"1000", "Border" : { "Width" : "5","Color" : "255,0,0", }, "ImageToolboxParams" : {"AlgorithmSetPath" :
              "c:/Software/CIB/_IPL_problems/2016_11_29/AlgorithmsSet_sample.xml","AlgorithmName" : "Invert", "ProfilePath1" :
              "c:/Software/CIB/_IPL_problems/2016_11_29/processing_profile.xml", },}, }

5.8. Overwrite

Specifies, must or not the library overwrite output images.

Syntax

Overwrite=<bool>
<bool>: true | false

Example

Overwrite=false

5.9. Threading

This property allows to control how many parallel threads should be used.

Syntax

Threading=”SelectionThreads:<number>, RenderThreads: <number>”

Example

Threading=”SelectionThreads: 3,RenderThreads: 3”

5.10. ErrorBehavior

This property controls what should happen in case of an error.

Syntax

ErrorBehavior=<behavior>
<behavior>: Stop| Continue | Fill

Stop:

In the case of a failed conversion, the procedure stops with a runtime error and the message “Error behaviour defined as ‘Stop’. Convertion was stopped.

Continue: (Default)

Fill:

Force creation of blob or image although convertion failed.

Example

ErrorBehavior=Continue

5.11. OriginPageCount

This property always refers to the source file and returns number of pages.

Syntax

OriginPageCount=<number>

Example

OriginPageCount=2

5.12. OriginMimeType

Returns the MimeType of the source file.

Syntax

OriginMimeType=<IANA MimeType>

Example

OriginMimeType=image/tiff

5.13. OriginImagePageInfo

Returns detailed meta information about the source files. For example resolution and size, color space and color depth, encoding and the according page selection.

Syntax

OriginImagePageInfo=
<ext>: tif|etc | 

Example

OriginImagePageInfo={"ColorDepth": 1, "ColorSpace": "Gray","Encoding": "Group4", "Height": 2327, "PageRange":"1-1", "ResolutionX": 200, "ResolutionY": 200,
"Width": 1647 } ;{ "ColorDepth": 8, "ColorSpace":"Gray", "Encoding": "JPEG", "Height":2348, "PageRange": "2-2", "ResolutionX": 200,"ResolutionY": 200, "Width": 1674 }

5.14. ResultFileNamePatterns

Information about the used patterns for filenames.

Syntax

ResultFileNamePatterns=<pattern>[;<pattern>]
<pattern>:<patternname>_###.<ext>
<ext>: png | jpg | tif |… 

Example

ResultFileNamePatterns=”pattern_png_%03d.png;pattern_jpeg_prev_%02d.jpg; pattern_jpeg_thumbs_%02d.jpg”

5.15. ResultCount

This JSON object describes the amount of pages that have been processed.

Syntax

ResultCount=<count per pagerange>[;<count per pagerange>]

Example

ResultCount{

                        "Count": 10,
                        "IsResultMultipage":false,
                        "PageRange": "1-10" 
}

5.16. ProcessingStatus

Provides information about the state of processing, which operation was or is conducted, in what area and if the operation was successful.

Syntax

ProcessingStatus=Operation:<operation>, PageRange: <pagerange>, RequiredRange:<requiredrange>, Status: 
                <status>
<operation>: Convertion | |
<pagerange>: <number> | <number>-<number>
<requiredrange>: <number> | <number>-<number>
<sstatus>: OK | NOK

Example

ProcessingStatus={Operation":"Convertion", "PageRange": "0","RequiredRange": "0", "Status": "OK" };{ "Operation": "Convertion", "PageRange":
                 "1", "RequiredRange": "1", "Status":"OK" } ;{ "Operation": "Convertion","PageRange": "0-1", "RequiredRange": "2","Status": "OK" }

Although the input for PageRange in the jobinterface starts with number 1 (e.g. “1-4”, the PageRange of the ProcessingStatus starts with (e.g. “0-3”).

Image Processing Library usage

Filters

For using filters you should perform next steps:

1. Include header for filters:

  #include "ImageProcessing_Filtration.h"

2. Create default utilities sets and copy them to new ProcessingProfile:

 UtilitiesSets sets = UtilitiesSetsCreator::CreateUtilitiesSets();
 ProcessingProfile profile;
 profile.m_Utilities = sets.utilities;
 profile.m_DebugUtilities = sets.debugUtilities;

3. Create profile parser and parse selected profile:

 shared_ptr<ProfileParser> profileParser(ProfileParserCreator::CreateProfileParser());
 profileParser->ParseProfile(profileFileName, profile);

4. Load image for processing:

  RasterImage imageToProcess;
  shared_ptr< IPL::IOStreams::ImageDecoder > imageDecoder(
  IPL::IOStreams::ImageDecodersFabric::CreateImageDecoder());
  IPL::IOStreams::InputStream inputStream(inputFileName);
  imageToProcess = imageDecoder->decode(inputStream);

5. Create default filter creator (fabric):

  shared_ptr<ImageFilterCreator> filterCreator(
  ImageFilterCreatorsSelector::SelectImageFilterCreator() );

6. Create filters:

  int binarizationFilterType = IFK_SAVOULA_BINARIZER;
  int noiseFilterType =IFK_HUANG_BILATERAL_FILTER;
  shared_ptr<ImageFilter> binarizationFilter(
  filterCreator->CreateFilter(binarizationFilterType, profile) );
  shared_ptr<ImageFilter> noiseFilter(
  filterCreator->CreateFilter(noiseFilterType, profile) );

7. Process image:

  RasterImage processedImage;
  processedImage = binarizationFilter->apply(noiseFilter->apply(imageToProcess));

8. Show processing result:

  profile.m_DebugUtilities->ShowImage("Window Title for Processed Image", processedImage);

9. And store it:

  
shared_ptr< IPL::IOStreams::ImageEncoder > encoder(
             IPL::IOStreams::ImageEncodersFabric::CreateImageEncoder(IPL::IOStreams::IEK_JPG_ENCODER));
IPL::IOStreams::OutputStream outputStream(inputFileName + ".proc" + ".jpg");
encoder->encode(processedImage, outputStream);

 Look for sample FiltrationSample() in ImageProcessing_sample.cpp.

Segmenters

For using segmenters you should perform next steps:

1. Include header for segmenters:

  #include "ImageProcessing_Segmentation.h"

2. Create default utilities sets and copy them to new ProcessingProfile:

  UtilitiesSets sets = UtilitiesSetsCreator::CreateUtilitiesSets();
  ProcessingProfile profile;
  profile.m_Utilities = sets.utilities;
  profile.m_DebugUtilities = sets.debugUtilities;

3. Create profile parser and parse selected profile:

  shared_ptr<ProfileParser> profileParser(ProfileParserCreator::CreateProfileParser());
  profileParser->ParseProfile(profileFileName, profile);

4. Load image for processing and create array for results:

  RasterImage imageToProcess;
  RasterImageArray results;
  profile.m_DebugUtilities->ReadImageFromFile(inputFileName, imageToProcess);

5. Create default segmenter creator (fabric)

  shared_ptr<ImageSegmenterCreator> segmenter_creator(
  ImageSegmenterCreatorsSelector::SelectImageSegmenterCreator() );

6. Create segmenter:

  int segmenterType = ISK_RANGES_SEGMENTER;
  shared_ptr<ImageSegmenter> segmenter(
  segmenter_creator->CreateSegmenter(segmenterType, profile));

7. Process image:

  segmenter->segment(imageToProcess, results);

8. Save processing result to files:

  for (size_t imageNo = 0; imageNo < results.size(); ++imageNo) {
       profile.m_DebugUtilities->WriteImageToFile(results[imageNo],
         inputFileName + ".layer_" + std::to_string(imageNo) + ".jpg");
  }
or
  shared_ptr<IPL::IOStreams::ImageEncoder> encoder(
    IPL::IOStreams::ImageEncodersFabric::CreateImageEncoder(IPL::IOStreams::IEK_JPG_ENCODER));
  for (size_t imageNo = 0; imageNo < results.size(); ++imageNo) {       
    IPL::IOStreams::OutputStream outoutStream(
         inputFileName + ".layer_" + std::to_string(imageNo) + ".jpg");   
    encoder->encode(results[imageNo], outoutStream);
  }

 Look for sample SegmentationSample() in ImageProcessing_sample.cpp.

User defined filters[edit | edit source]

For developing new filter (or segmenter) you should perform next steps:

1. Include header for filters:

  #include "ImageProcessing_Filtration.h"

2. Create new class which inherits class ImageFilter (or ImageSegmenter for segmenters).

3. Create implementations of methods for new class:

  virtual bool Initialize(ProcessingProfile& profile);
  virtual RasterImage::RasterImagePtr apply(RasterImage& image) final;

For using of new filter (or segmenter) you should perform next steps:

1. Create default filter creator (fabric):

  shared_ptr<ImageFilterCreator> filterCreator(
  ImageFilterCreatorsSelector::SelectImageFilterCreator() );

2. Create lambda for new filter creation:

  ImageFilterCreator::ImageFilterCreationProcedure newFilterCreationProcedure =   
[](ProcessingProfile& profile) -> ImageFilter*
{    
return new UserDefinedImageFilter(profile);  
};

3. Add new filter to fabric:

  uint32_t newFilterKey = filterCreator->AddFilter(newFilterCreationProcedure);

4. Create profile and set processing parameters

  ProcessingProfile profile;
  profile["UserDefinedImageFilter/Parameter1"] = "1";

5. Create instance of user defined filter

  shared_ptr<ImageFilter> userDefinedFilter(
  filterCreator->CreateFilter(newFilterKey, profile) );

6. Process image by filters

  
processedImage = userDefinedFilter->apply(noiseFilter->apply(imageToProcess)); Look for sample class
UserDefinedImageFilter and UserDefinedImageFilter_sample() in ImageProcessing_sample.cpp.

Usage of DLL interface

CIB Job interface

Functions of this interface described in IPL_Interface.h.

Interface for doXisafe

For usage of the library from doXisafe ScenarioExecuter is available. It can be used by calling function CibImageProcessingRunScenario() declared in IPL_ScenarioExecuter_Interface.h.

Improvement functionality using plugins

IPL functionality can be improved by external module which are separated dynamically loading libraries.

Plugin interface

For developing IPL plugin you should implement functions declared in file IPL_Plugin_Interface.h as .dll (for Windows) or .so (for Linux):

  PluginInterfaceVersionType GetPluginInterfaceVersion();
  bool InitializePlugin(const std::string& pluginParametersString);
  RasterImage* RunPlugin(RasterImage& inputImage);
  bool DestroyPlugin();

Also library IPL_Plugin_Base.lib should be linked to your plugin.

Plugin usage

For using of plugin special filter was implemented. Key of this filter is IPL::Filtration::IFK_PLUGIN_FILTER.

Plugin file and corresponding processing parameters should be defined in processing profile

  <ExternalFilter>
      <PluginFileName>IPL_Plugin_sample.dll</PluginFileName>
       <PluginParametersString>1</PluginParametersString>
  </ExternalFilter>

or manually by correction of properties map

  
profile["PluginFileName/PluginParametersString"] = "0";

Processing profiles

Processing profiles are used for describing of processing parameters for implemented filters. Profile is a file in XML format with next structure (order of descriptions and filters/segmenters parameters does not matter):

  
<ProcessingProfile>     
  <Filter1>       
    <Filter1_Parameter1>Filter1_Parameter1_Value</Filter1_Parameter1>      
    ...       
    <Filter1_ParameterN>Filter1_ParameterN_Value</Filter1_ParameterN>   
 </Filter1>     
 ...      
 <FilterM>          
    <FilterM_Parameter1>FilterM_Parameter1_Value</FilterM_Parameter1>       
    ...       
    <FilterM_ParameterK>FilterM_ParameterK_Value</FilterM_ParameterK>   
 </FilterM>     
 <Merger1>          
   <Merger1_Parameter1>Merger1_Parameter1_Value</Merger1_Parameter1>       
   ...        
   <Merger1_ParameterY>Merger1_ParameterY_Value</Merger1_ParameterY>    
 </Merger1>     
 ...     
 <MergerL>         
   <MergerL_Parameter1>MergerL_Parameter1_Value</MergerL_Parameter1>       
   ...         
   <MergerL_ParameterX>MergerL_ParameterX_Value</MergerL_ParameterX>     
 </MergerM>    
 <Segmenter1>       
   <Segmenter1_Parameter1>Segmenter1_Parameter1_Value</Segmenter1_Parameter1>
   ...      
   <Segmenter1_ParameterY>Segmenter1_ParameterY_Value</Segmenter1_ParameterY>    
 </Segmenter1>      
 ...    
 <SegmenterL>         
   <SegmenterL_Parameter1>SegmenterL_Parameter1_Value</SegmenterL_Parameter1>        
   ...
   <SegmenterL_ParameterX>SegmenterL_ParameterX_Value</SegmenterL_ParameterX>   
 </SegmenterM> 
</ProcessingProfile>

6. Image processing algorithms: Filters

Filters are divided into: Affine transform filters, Binarization filters, Modification filters, Noise Filters, Special filters.

6.1. Affine transform filters

CroppingFilter

This filter allows cropping of defined area of original image. Area for cropping can be defined in two ways:

• By separated borders,
• By single line.

In both cases you should keep in mind coordinate system of image. Pixel with coordinates (0;0) is located in top left corner of image:

For example definitions

<ProcessingProfile>
...
 <CroppingFilter>  
 <IsResizingToSourceSizeRequired>1</IsResizingToSourceSizeRequired>   
 <CroppingArea>100;500;100;500</CroppingArea> 
</CroppingFilter>
...
</ProcessingProfile>
<ProcessingProfile>
... 
<CroppingFilter>   
 <IsResizingToSourceSizeRequired>1</IsResizingToSourceSizeRequired>  
 <Top>100</Top>  
 <Bottom>500</Bottom>
 <Left>100</Left>
 <Right>500</Right>
</CroppingFilter>
...
</ProcessingProfile>

are equal.

Also filter can be used for cropping quadrangular areas. In this case user should define area as set of four clockwise points in format x0;y0;x1;y1;x2;y2;x3;y3:

<ProcessingProfile>
...
<CroppingFilter> 
 <IsResizingToSourceSizeRequired>1</IsResizingToSourceSizeRequired>
 <CroppingArea>100;100;500;250;500;500;150;500</CroppingArea>
</CroppingFilter>
...
</ProcessingProfile>

In addition with usage of flag IsResizingToSourceSizeRequired (if it is not equal 0) result of cropping can be resized to size of original image.

Also you can define background color:

<BackgroundColor>0;0;0</BackgroundColor>

DocumentCropper

The filter auto-crops an input image.

This algorithm has no parameters.

Example

<step command="ipl"expected-result-code="0" name="ImageProcessingStep_page_1” timeout="0">
<properties>
<property name="WorkSpace">.\</property>
<property name="OperationName">DocumentCropper</property>
<property name="OutputFileName">cropedImage.png</property>
<property name="OutputDirectory">.\</property>
<property name="InputFilename">inputs\Versatel.png</property>
<property name="TraceFilename">file.log</property>
</properties>
</step>

DocumentDeskewer

The filter deskews an input image.

Parameters:

<step command="ipl" expected-result-code="0" name="ImageProcessingStep_page_1” timeout="0">
<properties>
<property name="WorkSpace">.\</property>
<property name="OperationName">DocumentDeskewer</property>
<property name="OutputFileName">deskewedImage.png</property>
<property name="OutputDirectory">.\</property>
<property name="InputFilename">inputs\Versatel.png</property>
<property name="TraceFilename">file.log</property>
</properties>
</step>

FlippingFilter

Flips an input image.

Parameters:

ResizeFilter

Image resizing filter.

Parameters of image resizing can be set in two ways:

By scaling. In this case we should set parameter "ResizeFilter/ResizingParameter" to required scale value:

<ProcessingProfile>
...
 <ResizeFilter> 
  <ResizingParameter>0.5</ResizingParameter>
</ResizeFilter>
...
\endcod
</ProcessingProfile>e
-# By defining of target size. In this case we should set parameter 
"ResizeFilter/ResizingParameter"
to required size value:
\code{.xml}
<ProcessingProfile>
...
<ResizeFilter>  
 <ResizingParameter>W;H</ResizingParameter> 
</ResizeFilter>
...
</ProcessingProfile>

where W is target width, H – height. If W or H is equal to 0 then this value will be calculated depends on other size and source image geometry. For example:

<ProcessingProfile>
... 
<ResizeFilter>
    <ResizingParameter>0;1000</ResizingParameter>
</ResizeFilter>
...
</ProcessingProfile>

In this case width will be calculated:

height = 1000;

width = sourceImage.Width * (height / sourceImage.Height);

If W or H is less than 0 or empty then corresponding size will not be changed. For example samples

<ProcessingProfile>
... 
<ResizeFilter>   
<ResizingParameter>-1;1000</ResizingParameter> 
</ResizeFilter>
...
</ProcessingProfile>

and

<ProcessingProfile>
... 
<ResizeFilter>  
<ResizingParameter>;1000</ResizingParameter>
</ResizeFilter>
...
</ProcessingProfile>

are equal and create image with width <sourceImage.Width>x1000 pixels.

RotationFilter

Interface for image clockwise rotation filter.

DocumentCleaner

The DocumentCleaner gets an input image and tries to eliminate fog and uneven light so that the output looks almost like a scan.

Parameters of this filter can be described in profile:

<ProcessingProfile>
... 
<DocumentCleaner>   
  <BrightSpots>0</BrightSpots>
</DocumentCleaner>
...
</ProcessingProfile>

RotateAndMirrorFilter

Filter for rotation and image mirroring.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<RotateAndMirrorFilter>  
  <Degrees>0</Degrees> 
  <Mirror>false</Mirror>
</RotateAndMirrorFilter>
...
</ProcessingProfile>

6.2. Binarization filters

The following chapters explain each filter. Usually there is also an example given. The input image for the filters is the following picture:

COCOCLUST_Binarizer

Implementation of COCOCLUST binarization.

This binarizator is described in the publication Kasar, T. and Ramakrishnan, A.G., 2009. COCOCLUST: Contour-based color clustering for robust binarization of colored text. Proc. The Third CBDAR, pp.11-17.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<COCOCLUST_Binarizer> 
 <Threshold>45</Threshold>   
 <CLAHEClipLimit>3.0</CLAHEClipLimit>  
 <GaussianBlurKernelSize>19</GaussianBlurKernelSize> 
 <CannyUpperThresholdCoeff>0.15</CannyUpperThresholdCoeff> 
 <CannyLowerThresholdCoeff>0.05</CannyLowerThresholdCoeff> 
 <CannyMorphIters>4</CannyMorphIters>
</COCOCLUST_Binarizer>
...

</ProcessingProfile>

FBCITB_Binarizer

Implementation of FBCITB binarization.

This binarizator is described in Kasar, T., Kumar, J. and Ramakrishnan, A.G., 2007, September. Font and background color independent text binarization. In Second international workshop on camera-based document analysis and recognition (pp. 3-9).

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<FBCITB_Binarizer>  
 <UseCanny>1</UseCanny>  
 <UseVariancesMap>0</UseVariancesMap> 
 <VarianceMapThreshold>200</VarianceMapThreshold> 
 <CLAHEClipLimit>2.0</CLAHEClipLimit>  
 <GaussianBlurKernelSize>9</GaussianBlurKernelSize>   
 <CannyUpperThresholdCoeff>0.6</CannyUpperThresholdCoeff>  
 <CannyLowerThresholdCoeff>0.4</CannyLowerThresholdCoeff>  
 <BoundingRectangleMaxArea>0.3</BoundingRectangleMaxArea>
</FBCITB_Binarizer>
...
</ProcessingProfile>

FengBinarizerFilter

Implementation of Feng binarization.

This binarizator is described in Khurshid, K., Siddiqi, I., Faure, C. and Vincent, N., 2009, January. Comparison of Niblack inspired binarization methods for ancient documents. In Document Recognition and Retrieval XVI (Vol. 7247, p. 72470U). International Society for Optics and Photonics.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<FengBinarizerFilter>
 <WindowSize>21</WindowSize>
 <Alpha1>0.7</Alpha1>
 <k1>0.2</k1>
 <k2>0.03</k2>
 <Gamma>2.0</Gamma>
 <MorphIterationCount>2</MorphIterationCount>
</FengBinarizerFilter>
...
</ProcessingProfile>

GaussianBlurFilter

In image processing, a Gaussian blur is the result of blurring an image by a Gaussian function. This function used to reduce Image noise and detail.

Parameters of this filter can be described in profile:

Please note that “KernelSize” only accept odd numbers!

<?xml version="1.0" encoding="utf-8"?>
<algorithms>
<algorithm name="ImageManipulation" TracingEnabled="false" inputs="in1=$1" outputs="out4=$1">
<layer>
<processor params="KernelSize:21" type="GaussianBlurFilter" outputs="out4" inputs="in1" />
</layer>
</algorithm>
</algorithms>   

LocalOtsuBinarizer

Implementation of Otsu binarization algorithm.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <LocalOtsuBinarizer>
  <MaxValue>255.0</MaxValue>
  <CLAHEClipLimit>0.0</CLAHEClipLimit> 
  <GaussianBlurKernelSize>19</GaussianBlurKernelSize>  
  <CannyUpperThresholdCoeff>0.15</CannyUpperThresholdCoeff>
  <CannyLowerThresholdCoeff>0.01</CannyLowerThresholdCoeff>  
  <CannyMorphIters>1</CannyMorphIters>
  </LocalOtsuBinarizer>
...
</ProcessingProfile>

MedianBlurAGTFilter

Median blur + Adaptive Gaussian threshold filter.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <MedianBlurAGTFilter>
  <KernelSize>3</KernelSize>  
  <MaxValue>255.0</MaxValue> 
  <BlockSize>5</BlockSize>
  <Shift>0.0</Shift>
 </MedianBlurAGTFilter>
...
</ProcessingProfile>

MedianBlurATFilter

Median blur + Adaptive threshold filter.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <MedianBlurATFilter>  
  <KernelSize>3</KernelSize>
  <MaxValue>255.0</MaxValue> 
  <BlockSize>5</BlockSize> 
  <Shift>0.0</Shift>
 </MedianBlurATFilter>
...
</ProcessingProfile>

MedianBlurGATFilter

Gaussian blur + Adaptive threshold filter.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <MedianBlurGATFilter>
  <KernelSize>3</KernelSize>
  <SigmaX>10.0</SigmaX>
  <SigmaY>0.0</SigmaY>
  <MaxValue>255.0</MaxValue>
  <BlockSize>5</BlockSize>
  <Shift>0.0</Shift>
 </MedianBlurGATFilter>
...
</ProcessingProfile>

NativeAdaptiveBinarizer

Native adaptive thresholding.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <NativeAdaptiveBinarizer>  
  <IsGaussianBlurReqiured>0</IsGaussianBlurReqiured> 
  <MedianBlurKernelSize>5</MedianBlurKernelSize>
  <GaussianBlurKernelSize>3</GaussianBlurKernelSize>
  <GaussianBlurSigma>150</GaussianBlurSigma
  <IsAdaptiveThresholdCalculatedByGaussian>1</IsAdaptiveThresholdCalculatedByGaussian>
  <AdaptiveThresholdingMaxValue>255.0</AdaptiveThresholdingMaxValue>
  <AdaptiveThresholdingBlockSize>19</AdaptiveThresholdingBlockSize>
  <AdaptiveThresholdingShift>8</AdaptiveThresholdingShift>
  <BilateralFilterBlockSize>5</BilateralFilterBlockSize>
  <BilateralFilterColorSigma>50.0</BilateralFilterColorSigma>
  <BilateralFilterSpaceSigma>50.0</BilateralFilterSpaceSigma>
 </NativeAdaptiveBinarizer>
...
</ProcessingProfile>

NiblackBinarizerFilter

Implementation of Niblack binarization.

This binarizator is described in Khurshid, K., Siddiqi, I., Faure, C. and Vincent, N., 2009, January. Comparison of Niblack inspired binarization methods for ancient documents. In Document Recognition and Retrieval XVI (Vol. 7247, p. 72470U). International Society for Optics and Photonics.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 
<NiblackBinarizerFilter>  
 <WindowSize>101</WindowSize> 
 <ThresholdCoefficient>0.01</ThresholdCoefficient> 
 <MorphIterationCount>2</MorphIterationCount>
 </NiblackBinarizerFilter>
...
</ProcessingProfile>

NICKBinarizerFilter

Implementation of NICK binarization.

This binarizator is described in Khurshid, K., Siddiqi, I., Faure, C. and Vincent, N., 2009, January. Comparison of Niblack inspired binarization methods for ancient documents. In Document Recognition and Retrieval XVI (Vol. 7247, p. 72470U). International Society for Optics and Photonics.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<NICKBinarizerFilter>  
 <WindowSize>101</WindowSize>  
 <ThresholdCoefficient>0.01</ThresholdCoefficient>
 <MorphIterationCount>2</MorphIterationCount>
 </NICKBinarizerFilter>
...
</ProcessingProfile>

OtsuBinarizer

Interface for Otsu binarization algorithm.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <OtsuBinarizer> 
  <MaxValue>255.0</MaxValue>
  </OtsuBinarizer>
...
</ProcessingProfile>

PureAdaptiveGaussianThresholdingFilter

Pure adaptive Gaussian thresholding filter.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <PureAdaptiveGaussianThresholdingFilter> 
  <MaxValue>255.0</MaxValue>
  <BlockSize>5</BlockSize> 
  <Shift>0.0</Shift>
 </PureAdaptiveGaussianThresholdingFilter>
...
</ProcessingProfile>

PureAdaptiveThresholdingFilter

Pure adaptive thresholding filter.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <PureAdaptiveThresholdingFilter> 
  <MaxValue>255.0</MaxValue>
  <BlockSize>5</BlockSize>
  <Shift>0.0</Shift>
 </PureAdaptiveThresholdingFilter>
...
</ProcessingProfile>

SauvolaBinarizerFilter

Implementation of Sauvola binarization.

This binarizator is described in Shafait, F., Keysers, D. and Breuel, T.M., 2008, January. Efficient implementation of local adaptive thresholding techniques using integral images. In Document recognition and retrieval XV (Vol. 6815, p. 681510). International Society for Optics and Photonics.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<SauvolaBinarizerFilter> 
 <WindowSize>101</WindowSize>  
 <ThresholdCoefficient>0.01</ThresholdCoefficient>
 <MorphIterationCount>2</MorphIterationCount>
</SauvolaBinarizerFilter>
...
</ProcessingProfile>

ThresholdFilter

Pure thresholding binarizator.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <ThresholdFilter>
   
<ThresholdValues>100;128;90</ThresholdValues>
 </ThresholdFilter>
...
</ProcessingProfile>

Also we can define equal value of thresholds for each channel:

<ProcessingProfile>

...

 <ThresholdFilter>

    <ThresholdValues>128</ThresholdValues>

 </ThresholdFilter>

...

</ProcessingProfile>

WolfJolionBinarizerFilter

Implementation of WolfJolion binarization.

This binarizator is described in Khurshid, K., Siddiqi, I., Faure, C. and Vincent, N., 2009, January. Comparison of Niblack inspired binarization methods for ancient documents. In Document Recognition and Retrieval XVI (Vol. 7247, p. 72470U). International Society for Optics and Photonics.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 
<WolfJolionBinarizerFilter> 
 <WindowSize>101</WindowSize> 
 <ThresholdCoefficient>0.01</ThresholdCoefficient>
 <MorphIterationCount>2</MorphIterationCount>
 </WolfJolionBinarizerFilter>
...
</ProcessingProfile>

6.3. Modification filters

AlphaMaskFilter

Filter for alpha channel thresholding.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <AlphaMaskFilter>
  <Threshold>128</Threshold>
  <IsInvertedMaskRequired>0</IsInvertedMaskRequired> 
  <IsBinaryMaskRequired>1</IsBinaryMaskRequired>
  </AlphaMaskFilter>
...
</ProcessingProfile>

AutoInvertFilter

Implementation of auto inversion procedure.

Image will be inverted if a number of non-black point is greater or equal than half of image area.

This filter have no parameters.

CleanBackgroundToWhiteFilter

The filter cleans background with making all near-white colors true white color.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <CleanBackgroundToWhiteFilter>
  <gamma>1.0</gamma>
  <blackValue>70</blackValue>
  <whiteValue>170</whiteValue>
 </CleanBackgroundToWhiteFilter>
...
</ProcessingProfile>

CloseFilter

Morphology operation 'Close'.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <CloseFilter>  
  <IterationCount>1</IterationCount>
  </CloseFilter>
...
</ProcessingProfile>

ColorRangesFilter

Interface for color ranges filter.

This filter allows to extract from image all pixels with colors that fall within user defined ranges (or if it is required pixel that do not fall within these ranges). These ranges can be defined by definition of lower and upper range boundary:

<ProcessingProfile>
...
 
<ColorRangesFilter>
 <IsMaskOnlyRequired>0</IsMaskOnlyRequired>
 <IsInRangeColorsRequired>0</IsInRangeColorsRequired> 
 <LowerBoundaries>0,0,0;130,55,80</LowerBoundaries>
 <UpperBoundaries>128,128,128;209,190,190</UpperBoundaries>
 <RangeShapes>b(0, 0, 0, 127, 127, 127)</RangeShapes>
</ColorRangesFilter>
...
</ProcessingProfile>

Color ranges can be defined in two ways:

• by color values boundaries (LowerBoundaries and UpperBoundaries options)
• by definition of shapes in colorspace (RangeShapes option; this option has higher priority).

For color values boundaries usage we should define lower and upper values of color for each color range:

<LowerBoundaries>R1_L,G1_L,B1_L;R2_L,G2_L,B2_L</LowerBoundaries>
<UpperBoundaries>R1_U,G1_U,B1_U;R2_U,G2_U,B2_U</UpperBoundaries>

where R1_L – lower boundary of red color for the first range, R1_U – upper boundary of red color for the first range.

The same results can be obtained using the second way: we can define corresponding boxes (rectangular shapes):

<RangeShapes>b(R1_L,G1_L,B1_L, R1_U,G1_U,B1_U);b(R2_L,G2_L,B2_L, R2_U,G2_U,B2_U)</RangeShapes>

Also spherical shapes can be used:

<RangeShapes>s(centerR, centerG, centerB, diameter)</RangeShapes>

There is special way for spherical shapes usage: we can use set of spheres to definition of extensional areas. For example, if we have following spheres set:

<RangeShapes>s{(5, 5, 5, 30) - (50, 50, 50, 40) - (100, 100, 50, 50) - (100, 100, 200, 25) - (200, 100, 50, 25)}</RangeShapes>

then we will obtain next 3D area for color selection (intermediate spheres parameters would be calculated automatically).

If IsMaskOnlyRequired is not equal 0 then we obtain only binary mask for defined color ranges.

ColorSpaceConvertionFilter

Interface of image color space convertion procedure.

This filter can be used for transforming current color space of image (this information can be obtained from input image) to user defined color space.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <ColorSpaceConvertionFilter>
  <TargetColorspace>HSV</TargetColorspace>
  <SourceColorspace>RGB</SourceColorspace>
 </ColorSpaceConvertionFilter>
...
</ProcessingProfile>

Currently implemented next conversions are:

• RGB -> GRAY
• RGB -> BGR
• RGB -> HSV
• RGB -> HLS
• RGB -> Lab
• RGB -> Luv
• RGB -> XYZ
• RGB -> YUV
• RGB -> YCrCb
• GRAY -> RGB
• BGR -> RGB
• HSV -> RGB
• HLS -> RGB
• Lab -> RGB
• Luv -> RGB
• XYZ -> RGB
• YUV -> RGB
• YCrCb -> RGB

ContrastEnhancementFilter

Contrast enhancement by CLAHE algorithm.

This filter enhance local contrast of image using algorithm CLAHE.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <ContrastEnhancementFilter>
  <CLAHEClipLimit>2.0</CLAHEClipLimit>
  </ContrastEnhancementFilter>
...
</ProcessingProfile>

DilateFilter

Morphology operation 'Dilate'.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <DilateFilter>
  <IterationCount>1</IterationCount>
 </DilateFilter>
...
</ProcessingProfile>

ErodeFilter

Morphology operation 'Erode'.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <ErodeFilter>
  <IterationCount>1</IterationCount>
 </ErodeFilter>
...
</ProcessingProfile>

GuoHallThinningFilter

Interface for thinning a binary image using Guo-Hall algorithm.

In details this algorithm is described in Kumar, H. and Kaur, P., 2011. A comparative study of iterative thinning algorithms for BMP images. International journal of computer Science and information Technologies (IJCSIT).

This filter has no parameters.

HistogramEqualizationFilter

This filter normalizes the image brightness and contrast by normalizing its histogram.

This filter have no parameters.

InvertFilter

Interface for image color inversion.
This filter has no parameters.

OpenFilter

Morphology operation 'Open'.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <OpenFilter>
  <IterationCount>1</IterationCount>
 </OpenFilter>
...
</ProcessingProfile>

ZhangSuenThinningFilter

ColorRangeConverter

Converts the color range of the image.

This filter creates the new image using the equation:

dst(x,y) = src(x,y) * Multiplier + Shift

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <ColorRangeConverter>
  <Multiplier>1.0</Multiplier> 
  <Shift>0.0</Shift>
 </ColorRangeConverter>
...
</ProcessingProfile>

BalanceWhiteFilter

Filter for white balance correction.

Equation that describes the output image:

dst(x,y) = WhiteBalance(src(x,y)) * Multiplier + Shift

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <BalanceWhiteFilter>
  <SaturationThreshold>0.9</SaturationThreshold>
  <Multiplier>1.8</Multiplier>
  <Shift>0.0</Shift>
 </BalanceWhiteFilter>
...
</ProcessingProfile>

ColorBalanceFilter

Filter for color balance correction.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <BalanceWhiteFilter>
  <ColorBalanceGamma>1.0</ColorBalanceGamma>
  <SaturationGamma>1.0</SaturationGamma>
 </BalanceWhiteFilter>
...
</ProcessingProfile>

ColorSwapFilter

Filter for color channels swap.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <ColorSwapFilter> 
  <SwapPercent>1</SwapPercent>
  <SwappedChannels>RB</SwappedChannels>
 </ColorSwapFilter>
...
</ProcessingProfile>

6.4. Noise filters

BilateralFilter

Implementation of generic bilateral filter.

This filter is described in Zhang, M., 2009. Bilateral filter in image processing.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 
<BilateralFilter>
 <KernelSize>5</KernelSize> 
 <ColorSigma>100</ColorSigma>
 <SpaceSigma>150</SpaceSigma>
</BilateralFilter>
...
</ProcessingProfile>

HuangBilateralFilter

Implementation of Huang bilateral filter.

This filter is described in Huang, Y.L. and Fuh, C.S., 2006. Noise reduction using enhanced bilateral filter. Images and Recognition, 12(4), pp.46-53.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 
<HuangBilateralFilter> 
 <KernelSize>5</KernelSize>
 <ColorSigma>100</ColorSigma>
 <SpaceSigma>150</SpaceSigma>
</HuangBilateralFilter>
...
</ProcessingProfile>

MedianBlurFilter

Interface for median blur filter.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
 <MedianBlurFilter>
  <KernelSize>5</KernelSize>
 </MedianBlurFilter>
...
</ProcessingProfile>

6.5. Special filters

CompositeFilter

Special filter for algorithms execution.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<CompositeFilter>
 <AlgorithmsSetFilename>../AlgorithmsSet_sample.xml</AlgorithmsSetFilename>
 <AlgorithmName>SimpleAlgorithm</AlgorithmName>
</CompositeFilter>
...
</ProcessingProfile>

ExternalFilter

Filter for plugin usage.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<ExternalFilter> 
 <PluginFileName>IPL_Plugin_sample.dll</PluginFileName> 
 <PluginParametersString>1</PluginParametersString>
</ExternalFilter>
...
</ProcessingProfile>

7. Image information

InfoGetBoundaries

This operation allows detection of boundaries of the main object in the image. Please find further information in CIB aiModule.

Parameters of this function:

Output properties: Boundaries, and Edges.

Boundaries: 8 numbers separated with semicolons (’;’) that all together represent the 4 corners of the bounding box.

Edges: Each edge consist of 4 numbers split with semicolons (‘;’). These numbers represent a line defined from border to border of the image in the format X0;Y0;X1;Y1.

Example

<job>
<steps> 
<step command="ipl">    
 <properties>      
  <property name="OperationName">InfoGetBoundaries</property>     
  <property name="Profile:InfoGetBoundaries/Algorithm">Ai</property>    
  <property name="Profile:InfoGetBoundaries/ModelFilename">document_v1.pb</property>    
  <property name="InputFilename">D:\PixelOne.jpg</property>    
  <property mode="out" name="InfoGetBoundaries/Boundaries"/>     
  <property mode="out" name="InfoGetBoundaries/Edges"/>   
 </properties> 
 </step>
</steps>

</job>             

Example (edges)

Request:

<property name="OperationName">InfoGetBoundaries</property> 
… 
<property mode="out" name="Profile:InfoGetBoundaries/Edges"/>

Output:

<property name="InfoGetBoundaries/Edges">
 
2550;0;2832;4048;2407;0;3048;4048;2263;0;3124;4048;2108;0;3191;4048;…
</property>

InfoGetExif

This operations extracts the Exif information from the input image.

There are no input parameters.

Output property: Exif.

8. Mergers

ImageChannelsMerger

Merger for joining individual images.

Parameters of this merger can be described in profile:

<ProcessingProfile>
...
 <ImageChannelsMerger>
 <TargetColorSpace>RGB</TargetColorSpace>
 </RDO_Segmenter>
...
</ProcessingProfile>

ColorQRMerger

Merger for joining individual images belonging to QR codes (experimental). Each input image represents a channel in the RGB color space.

There are no parameters for this merger, besides the images themselves. Maximum number of input images is limited to the range [1-3].

AlphaChannelMerger

Interface for tools for alpha channel merging.

Parameters of this merger can be described in profile:

<ProcessingProfile>
...
 <AlphaChannelMerger>
 <AlphaChannelValue>255</AlphaChannelValue>
 </AlphaChannelMerger>
...
</ProcessingProfile>

9. Segmenters

RangesSegmenter

Interface of image segmentation by color ranges.

This segmenter allows to extract from image to separated images all pixels with colors that fall within user defined ranges (or if it is required pixel that do not fall within these ranges). These ranges can be set by definition of lower and upper range boundaries:

<RangesSegmenter GetMasks="0">
<UsedColorSpace>RGB</UsedColorSpace>
<LowerBoundaries>0,0,0;130,55,80;219,200,130</LowerBoundaries>
<UpperBoundaries>128,128,128;209,190,190;255,255,191</UpperBoundaries>
</RangesSegmenter>

RDO_Segmenter

Segmenter for extracting layers from an image.

This segmentation procedure is based on article Cheng, H. and Bouman, C.A., 2001. Document compression using rate-distortion optimized segmentation. Journal of Electronic Imaging, 10(2), pp.460-475. But RDO_Segmenter is modification of algorithm from the article.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<RDO_Segmenter> 
<ImageBlockSize>8</ImageBlockSize> 
<IsTextBinaryMaskRequired>true</IsTextBinaryMaskRequired>
<ThreshHigh>150.0</ThreshHigh>
<ThreshLow>25.0</ThreshLow>
<ThreshMean>200.0</ThreshMean> 
<DCTCoeff>400</DCTCoeff>
<ThreshVariance>25</ThreshVariance>
<IsDownsamplingRequired>false</IsDownsamplingRequired> 
<RatioForeground>4.0</RatioForeground>
<RatioBackground>2.0</RatioBackground>
</RDO_Segmenter>
...
</ProcessingProfile>

ColorQRSegmenter

Segmenter for extracting layers from a QR RGB image (experimental). It only accepts 1 input image, and outputs 3 images, one per channel.

Parameters of this filter can be described in profile:

<ProcessingProfile>
...
<ColorQRSegmenter>
 <IsTextBinaryMaskRequired>true</IsTextBinaryMaskRequired>
</ColorQRSegmenter>
...
</ProcessingProfile>

SplitChannelsSegmenter

This segmenter allows to separate multichannel image to few grayscale images with one channel.

There are no parameters for this segmenter.

10. Image comparer

ComparePSNR

Compares two images and return the PSNR.

Output property: PSNR.

CompareDifference

Compares two images and return the difference.

Output property: Difference.

11. Algorithms sets

Algorithms sets are used for describing of complex image processing procedures. Algorithm can be described in two ways:

• sequence of image filters,
• directed graph of image processors.

<algorithms> 
...
      <algorithm name="SimpleAlgorithm">
          <filter type="HuangBilateralFilter" params="KernelSize:11" />          
<filter type="SavoulaBinarizerFilter" params="WindowSize:101|ThresholdCoefficient:0.015" />
          <filter type="OpenFilter" params="IterationCount:2" />
      </algorithm>  ...
  </algorithms>

This sample shows sequential filtration of single input image by three filters:

The second way allows usage of all processors (filters, segmenters, mergers). In this case we describe algorithm as set of processor blocks combined to layers and input/output of each block. For example:

  <algorithms> 
...>      
<algorithm name="GraphAlgorithm" inputs="in1=$1|in2=$2" outputs="l1p1o1=$1|l2p1o1=$2">         
<layer>
               <processor type="ColorRangesFilter" params="IsMaskOnlyRequired:0|IsInRangeColorsRequired:1|LowerBondaries:0,0,0|
UpperBondaries:96,96,96"
inputs="in1" outputs="l1p1o1" />
               <processortype="ColorSpaceConvertionFilter" params="TargetColorspace:GRAY" inputs="in2"
outputs="l1p2o1" />          
</layer>>          
<layer>
               <processor type="AlphaChannelMerger" params="AlphaChannelValue:128"
inputs="l1p1o1|l1p2o1" outputs="l2p1o1" />          
</layer>      
</algorithm> 
...  
</algorithms>

This sample shows processing of two files with further merging of preliminary processing results:

12. CIB aiModule

Description

Currently, CIB aiModule is only called via CIB image toolbox. Its functionality is to detect boundaries in order to auto crop images correctly.

The following libraries are required to run CIB aiModule:

Properties

The following properties are required for CIB image toolbox to call cropping with CIB aiModule:

Further supported input properties of CIB aiModule:

Output properties of CIB aiModule:

13. Returncodes

*This table contains errors of the IPL. Bold errorcodes indicate standard returncodes.