Active contour model driven by local histogram fitting energy

A nonparametric local region-based active contour driven by a local histogram fitting energy is presented. The energy is defined in terms of an evolving curve and two fitting histograms that approximate the distribution of object and background locally through a truncated Gaussian kernel. The kernel width for computing the fitting histograms should be different on different pixels, since the same kernel width applied may cause local minima of the energy. Three inequalities are introduced to determine whether larger kernel width should be considered. We do not assume any distributions in the presented method. The method therefore belongs to a nonparametric local region based active contour, and it can segment the regions whose distribution is hard to be predefined. Experimental results show desirable performances of our method. (C) 2013 Elsevier B.V. All rights reserved.