Table of content

New Features

Creation of the new Image Geometric transformation library
AVX 512 support

New Algorithms

Library	Algorithm	Comments
Color image operations	Color Space Conversion	convert a color image from a given color space to an other (with support for various color space, see Color models)
Classification image operations	Kernel K-means algorithm	classifies pixels of an image using Kernel k-means algorithm
Classification image operations	Kernel K-means assignment	assignment part of Kernel based KMeans clustering algorithm
Filtering image operations	Z Stack Focus Maximum	combine plans with maximum contrast in a 3d image to get a well contrasted 2d result
Filtering image operations	Z Stack Focus Ponderation	combine plans (ponderated by their contrast) in a 3d image to get a well contrasted 2d result
Global measure image operations	Global Entropy 2d	compute 2d image global entropy
Global measure image operations	Z ProjectionImg algorithm	measure of common statistics indicators in the image 3d (mean, max, etc.) computed along Z axis
Basic Morphological image operations	Union linear closing 2d	Union of closings calculated for several linear structuring element orientations
Basic Morphological image operations	Union linear closing 3d	Union of closings calculated for several linear structuring element orientations on 3d images
Basic Morphological image operations	Union linear opening 2d	Union of opening calculated for several linear structuring element orientations
Basic Morphological image operations	Union linear opening 2d	Union of opening calculated for several linear structuring element orientations on 3d images
Arithmetic image operations	Addition between images with different dimensions	add images with different dimensions
Arithmetic image operations	Division between images with different dimensions	divide images with different dimensions
Arithmetic image operations	Subtraction between images with different dimensions	subtract images with different dimensions
Arithmetic image operations	Multiplication between images with different dimensions	multiply images with different dimensions

New Shape Measures

New measures for Shape analysis library

Measure Type	Measure	Comments
Neighborhood	Distance to nearest barycenter 2d	computation of distance from shape barycenter to nearest neighbor shape barycenter for each shape
Neighborhood	Distance to nearest barycenter 3d	computation of distance from shape barycenter to nearest neighbor shape barycenter for each shape

New Low Level features

The parameter PolygApproxMaxDist is now set to 0.5 by default for polygon approximation. It avoids null area measure.

New C++ features

Library	Feature	Comments
IPSDKBaseShapeAnalysis	ipsdk::shape::analysis::applyArithmeticFormula	function allowing to apply an arithmetic formula on an existing measure set
IPSDKBaseShapeAnalysis	ipsdk::shape::analysis::applyLogicFormula	function allowing to apply a logic formula on an existing measure set
Image processing attributes	ipsdk::imaproc::attr::createBinaryLUT	function allowing to create an IntensityLut object dedicated to label to binary filtering
Image processing attributes	ipsdk::imaproc::attr::createLabelLUT	function allowing to create an IntensityLut object dedicated to label to label filtering with reindexing policy

New Python features

add better support for Kernel based objects
add extrema search functions
add gaussian curve Estimator
add extrema search function to Histogram class (see ipsdk::math::BaseHistogram)
add gaussian curve fitting to Histogram class (see ipsdk::math::BaseHistogram)
add KMeans, KernelKMeans and spectral clustering algorithms to Histogram class (see ipsdk::math::BaseHistogram)
add functions allowing to apply a formula on a computed measure set (resulting from a shape analysis). Computed formula can be arithmetical (PyIPSDK.applyArithmeticFormula) or logical (PyIPSDK.applyLogicFormula).
add functions allowing to create an IntensityLut object dedicated to :
- label to binary filtering (see PyIPSDK.createBinaryLUT)
- label to label filtering with reindexing policy (see PyIPSDK.createLabelLUT) These lut can be used in combination with logic formula described above.
IPSDK now supports python 3.7.

Behavior modifications

Modified Algorithms

Library	Algorithm	Comments
Color image operations	RgbToYuvImg	algorithm replaced by ColorConvertImg_Grp
Color image operations	YuvToRgbImg	algorithm replaced by ColorConvertImg_Grp
Advanced Morphological image operations	DilateLocalExtrema2dImg_Grp	InDilateFactor now accepts floating point values
Advanced Morphological image operations	DilateLocalExtrema3dImg_Grp	InDilateFactor now accepts floating point values
Advanced Morphological image operations	WatershedBinarySeparation2dImg_Grp	InDilateFactor now accepts floating point values
Advanced Morphological image operations	WatershedBinarySeparation3dImg_Grp	InDilateFactor now accepts floating point values
Advanced Morphological image operations	WatershedGreySeparation2dImg_Grp	InDilateFactor now accepts floating point values
Advanced Morphological image operations	WatershedGreySeparation3dImg_Grp	InDilateFactor now accepts floating point values
Shape analysis	ShapeFiltering2dImg_Grp	Add optional parameter allowing to specify custom measures definition for filtering (such as histogram quantile with target quantile)
Shape analysis	ShapeFiltering3dImg_Grp	Add optional parameter allowing to specify custom measures definition for filtering (such as histogram quantile with target quantile)
Basic Morphological image operations	DistanceMap2dImg_Grp	Possibility to specify the parameter to constrain the propagation
Basic Morphological image operations	DistanceMap3dImg_Grp	Possibility to specify the parameter to constrain the propagation
Advanced Morphological image operations	GenericSeededDistanceMap2dImg_Grp	Possibility to specify the parameter to constrain the propagation
Advanced Morphological image operations	GenericSeededDistanceMap3dImg_Grp	Possibility to specify the parameter to constrain the propagation
Advanced Morphological image operations	SeededDistanceMap2dImg_Grp	Possibility to specify the parameter to constrain the propagation
Filtering image operations	PatchBasedBilateral2dImg_Grp	Possibility to specify a single parameter for each plan in multislice case
Filtering image operations	PatchBasedBilateral3dImg_Grp	Possibility to specify a single parameter for each plan in multislice case

Modified Shape Measures

Measure Type	Measure	Comments
Inertia	Barycenter2d	Add support for calibration
Inertia	Inertia2d	Add support for calibration
Inertia	Barycenter3d	Add support for calibration
Inertia	Inertia3d	Add support for calibration

Modified Low Level features

Modified C++ features

Library	Feature	Comments
IPSDKImageFile library	ipsdk::image::file::saveTiffImageFile	Add possibility to compress saved image file (default is now Lzw compression)
IPSDKImageFile library	ipsdk::image::file::saveTiffImageFile	processing time optimization in case of RGB tiff image
IPSDKImageFile library	ipsdk::image::file::loadTiffImageFile	Add possibility to read 1 bit per pixel binary images
IPSDKImageFile library	ipsdk::image::file::loadTiffImageFile	processing time optimization in case of RGB tiff image

Bug Fixes

Algorithms Bug Fixes

Library	Algorithm	Comments
Shape analysis	Shape Analysis 2d	The geometric calibration of grey level images are now taken into account
Shape analysis	Shape Analysis 3d	The geometric calibration of grey level images are now taken into account
IPSDKImage library	Image geometry concepts in IPSDK	The image size is now coded on 64 bits (instead of 32), which allows to manage very large image buffers
IPSDKImageProcessing library	Output Image Initializer	The custom output initializer does not loose the geometric calibration data anymore
Image Geometric transformation	2d zoom	The geometric calibration is adapted to the zoom factors
Image Geometric transformation	3d zoom	The geometric calibration is adapted to the zoom factors

Low Level Bug Fixes

The file tbb.dll is now also installed in the Python binaries subdirectories. This is not exactly a bug but it avoids unwanted behaviour if another tbb.dll is installed in a system path.

Miscellaneous

Minimum CMake version : 3.13 because of Visual studio 2017 /FC option. For VS 2015 and before, the /FC option is disabled if not specified. From VS 2017, this option is enabled, which mofifies the behaviours of some macros. For instance, _ _ FILE _ _ yields a string without capital letter. This issue is managed by CMake for version >= 3.13.