Additive Local and Global Intensity based Active Contour Model for Inhomogeneous Image Segmentation

The involvement of image segmentation in many research fields has permitted the development of various efficient methods including the active contour based on level set formulation methods. There are some serious problems such as intensity inhomogeneity and re-initialization which exist in image segmentation and level set formulation. In the aim of overcoming these drawbacks, we propose an improved algorithm for inhomogeneous image segmentation with better computational time. The energy functional of the proposed method results from combining the local intensity information, global intensity information and some regularization factors. Firstly, the local intensity term is improved by introducing the general Gaussian as the kernel function which can accurately describe the image intensities within the pixels. Secondly, the global intensity term is based on a new scheme formulation that considers two intensity values for each region instead of one as in some existing algorithms. Moreover, the algorithm speed is boosted by eliminating the costly re-initialization procedure. Extensive experiments using various images have been carried out to illustrate the performance of the proposed method.