Image Segmentation Based on Two-Dimensional Inter-Class Cross Entropy and Chaos Optimization Algorithm

In this paper, a thresholding technique based on two-dimensional inter-class cross entropy and chaos optimization algorithm (COA) is presented. Firstly, adopt the cross entropy approach to describe the difference in pixel information between the object and the background in a two-dimensional histogram and to use maximum inter-class separation degree. Then apply improved COA to get the threshold value, which can get global solution with low computational load. Experimental results show the proposed approach is effective and gets competitive visual effects.