Gray-Level Image Segmentation Based on Markov Chain Monte Carlo
Wenqing Huang, Kebiao Zheng, Xiaokai Zhu · 2009
A new method called Markov chain Monte Carlo (MCMC) is proposed for image segmentation. The MCMC method mainly contains three aspects. Firstly, the image segmentation problem is formulated in a Bayesian statistical framework. Four types of gray-level image models are set up. Secondly, the solution space is decomposed into a union of many subspaces. Thirdly, ergodic Markov chains are designed to explore the solution space and sample the posterior probability. We test the MCMC algorithm on a wide variety of gray-level images and some results are shown in the paper.