Color image segmentation based on Bayesian framework and level set

Level set method is a power tool for tracking the curve evolution. From the view, we propose a novel color image segmentation method based on Bayesian and level set. Firstly, we regard the color information of each pixel as a stochastic variable that obeys certain probability distribution and deduce the segmentation model by Bayesian maximum a posteriori (MAP). Then we construct the corresponding energy function. Finally, we obtain the curve evolution equation based on multivariate normal distribution by variational method. We choose color remote sensing images and natural images to validate the proposed approach. The experimental results show that it is a feasible and effective approach to color image segmentation.