Code

Combination of piecewise-geodesic paths

The C++ source code for the algorithm building closed contours using combinations of piecewise-geodesic paths is available on GitHub.
Related papers are:

J. Mille, S. Bougleux and L.D. Cohen, Combination of piecewise-geodesic paths for interactive segmentation, International Journal of Computer Vision, 112(1):1-22, 2015
J. Mille, S. Bougleux and L.D. Cohen, Combination of paths for interactive segmentation, British Machine Vision Conference (BMVC), Bristol, England, 2013

which can be downloaded in PDF from the Publications page.

You can clone the deposit on GitHub or download a ZIP file of the current master revision at
https://github.com/julien-mille/comb-pgeo-paths
(last update : may 2015)

The code is provided with header files of libjpeg, libpng and zlib, as well as corresponding prebuilt .lib and .a library files for MSVC, MinGW and gcc for 32-bit x86 architecture only. If you wish to work under MacOS or with 64-bit versions of Windows or Linux (x86-64 architecture), you should rebuild the libraries on your own or install the package versions. You can follow the links to download them: libjpeg, libpng and zlib


Narrow band region-based active contour

The C++ source code for the narrow band region-based active contour model can be downloaded here. Related papers are:

J. Mille, Narrow band region-based active contours and surfaces for 2D and 3D segmentation, Computer Vision and Image Understanding, 113(9):946-965, 2009
J. Mille and L.D. Cohen, A local normal-based region term for active contours, 7th International Conference on Energy Minimization Methods in Computer Vision and Pattern Recognition (EMMCVPR), pages 168-181, Bonn, Germany, 2009

which can be downloaded in PDF from the Publications page.

Both 2D explicit (parametric) and implicit (binary level-set/region) implementations are provided. You may switch between the two models at the beginning of the main function. Note that 3D implementations are not available yet.
The source code can be compiled under Windows using Microsoft Visual C++ or MinGW, or under Linux using gcc. A .vcproj file for MSVC and a makefile for MinGW and gcc are provided in the archive.
Supported image formats are BMP, JPEG and PNG. Bitmap files support is natively implemented in the provided image class, so that no external library is needed for this particular format. JPEG files are supported thanks to libjpeg, whereas PNG format support relies on libpng and zlib.

The project is available in the following archive: activecontour.tar.gz (last update : november 2013)
The code is provided with header files of libjpeg, libpng and zlib, as well as corresponding prebuilt .lib and .a library files for MSVC, MinGW and gcc for 32-bit x86 architecture only. If you wish to work under MacOS or with 64-bit versions of Windows or Linux (x86-64 architecture), you should rebuild the libraries on your own. You can follow the links to download them: libjpeg, libpng and zlib