Active contour model via local and global intensity information for image segmentation

Abstract

Active contour models (ACM) have been proven to be the most promising model in solving the different problems encountered in image segmentation. This paper proposes a new region-based active contour model for level set formulation in which the energy function is formulated using both local and global intensity fitting terms. The generalized Gaussian distribution has been used as the kernel function of the local binary fitting information. The evolution equation consists of three terms: the global term, the local term and the regularization term. We have introduced the Laplace operator into the regularization term to regularize the level set function during its evolution process, which efficiently eliminates the costly re-initialization procedure. Due to the definition of local image intensities, the proposed model is able to deal with intensity inhomogeneity. The proposed model has been successfully applied to many synthetic and real-world images and the experimental results clearly show some improvement on both efficiency and accuracy compared with some popular methods.