Abstract: We present a kind of image segmentation based on the region of color space and LBP algorithm. It uses color and LBP histogram as the descriptors of the region and quantify the HSV color space, using the Uniform pattern of LBP algorithm to describe most of the texture features. It reduces the number of regional dimension feature vectors effectively and improves the execution efficiency of regional similarity calculation. The results have a good consistency with visual human eyes and more better than the MSRM algorithm in detail and execution efficiency.