Kernel graph cut image segmentation

Kernel graph cut segmentation according to the formulation in M. Ben Salah et al., IEEE TIP, 2011.

This code implements multi-region graph cut image segmentation according
to the kernel-mapping formulation in M. Ben Salah, A. Mitiche, and
I. Ben Ayed, Multiregion Image Segmentation by Parametric Kernel Graph
Cuts, IEEE Transactions on Image Processing, 20(2): 545-557 (2011).

The code uses Veksler, Boykov, Zabih and Kolmogorov’s implementation
of the Graph Cut algorithm. Written in C++, the graph cut algorithm comes
bundled with a MATLAB wrapper by Shai Bagon (Weizmann), which has to be downloaded from the following link (Matlab Wrapper for Graph Cuts):

http://www.wisdom.weizmann.ac.il/~bagon/matlab_code/GCmex1.9.tar.gz

The kernel-mapping part was implemented in MATLAB by M. Ben Salah (University of Alberta). If you use this code, please cite the papers mentioned in the
accompanying bib file (citations.bib).

The kernel-mapping formulation can handle various type of images, including
color photographs as well as data corrupted by a strong multiplicative
noise as in remote sensing synthetic aperture radar (SAR) or medical imaging
ultrasound. It is an efficient and flexible alternative to explicit modeling
of imaging noise using standard distributions (e.g., Gamma, Raleigh, Exponential,
Gaussian, Weibull, etc.).

Complete details on usage and compilation can be found in the enclosed
pdf file (Readme.pdf).

This code was tested on the following versions of MATLAB and C++:

MATLAB Version: 7.12.0.635 (R2011a) for 32-bit wrapper
Microsoft Visual C++ 2010 Express