A New Statistical Active Contour Model for Noisy Image Segmentation

This paper addresses the segmentation problem in noisy image based on Fast Edge Integration (FEI) method in active contour model (ACM) and proposes a new statistical active contour model (SACM). Two modifications are performed in FEI method. First, in order to handle noisy images, maximum log-likelihood estimation is used to replace the minimal variance term proposed by Chan and Vese. Second, a penalising term is employed to replace the time consuming re-initialization process. The proposed SACM is evaluated and compared with the existing ACM-based algorithms in terms of segmentation results and computational time. The proposed SACM outperforms existing methods and requires much less computational time.