Simple tools used to evaluate the quality of image segmentation algorithms according to a groundtruth image.
Both tools work only on 2D and 3D images with an unsigned char pixel type. All types of images supported by ITK are accepted. You can also specify a directory, in which cases all bmp, jpeg or png images in it will be sorted and loaded in order to build a 3D image.
Both tools print theirs results to stdout using a YAML format.
Computes the Hausdorff distance between two binary images.
$ ./HausdorffDistanceComputer -h
Usage: ./HausdorffDistanceComputer segmentation groundtruth
See http://www.itk.org/Doxygen/html/classitk_1_1HausdorffDistanceImageFilter.html for more informations about the Hausdorff distance.
Compute overlap measures between two labelled images (where a pixel value represents the pixel's class). It supports from 2 to 256 classes.
This tool accepts an optional mask image that will restrict the region of interest.
$ ./LabelOverlapComputer -h
Usage: ./LabelOverlapComputer segmentation groundtruth [mask]
See http://www.itk.org/Doxygen/html/classitk_1_1LabelOverlapMeasuresImageFilter.html for more informations about overlap measures.
Run CMake, and during the "configure" step, set the ITK_DIR variable to the lib/cmake/ITK-X.Y/ directory of your ITK installation.
The itkMaskedLabelOverlapMeasuresImageFilter.h and itkMaskedLabelOverlapMeasuresImageFilter.hxx files are licensed under the Apache License, Version 2.0. See the LICENSE-Apache-2.0.txt file for more details.
Other files are released under MIT License. See the LICENSE-MIT.txt file for more details.