A method of dunhuang frescos segmentation based on Markov random field and Graph cut

摘要

This paper combined the dependence of statistical features of pixels and the low-level features of image to complete image modeling. Using interactive image segmentation principle, through the posterior probability of maximum labelling field to obtain global energy function. Applying Graph Cut method to minimize the energy function and calculating optimal segmentation labeling of the whole image. Then, this paper apply it into the segmentation of dunhuang fresco, and we find the result is conspicuously better than typical grab cut algorithm, consequently, the effect of this paper is proved.