Comparison of Region Based Active Contour Models

Image segmentation is a fundamental task in image analysis responsible for partitioning an image into multiple sub-regions based on a desired feature. Active contour models (ACMs) have been widely used as attractive image segmentation methods because they always can achieve sub-pixel accuracy, provide closed and smooth contours/surfaces and can also be treated as automatic segmentation while the kernel-based edge detection methods, e.g. Sobel edge detectors, often produce discontinuous boundaries. This paper describes the comparative analysis of the three region based ACMs namely active contour without edges model (Standard Chan-Vese (C-V) Model), localized active contour without edges model and Local Binary Fitting (LBF) model. The experimental results shows that active contour without edges model fails to segment intensity inhomogeneous image whereas localized active contour without edges model and LBF model can successfully handle intensity inhomogeneity. However, the final output of localized active contour without edges and LBF models strongly depend on proper selection of localization radius. Based on time consumption, LBF is most time efficient as it doesn’t require time inefficient re-initialization for the evolving level set function.