IMAGE SEGMENTATION MODEL BASED ON CROSS ENTROPY AND IMPROVED SPARROW SEARCH ALGORITHM

Traditional image segmentation method based on entropy criteria uses the exhaustive method to search the segmentation thresholds, which has the shortage of high computational cost and poor segmentation efficiency. In order to solve this problem, this paper proposes an image segmentation method based on cross entropy and improved sparrow search algorithm. In order to improve the optimization accuracy and the optimization speed of standard sparrow search algorithm, we used the oppositelearning mechanism to conduct population initialization and improve the initial population structure, which could diversify the population and promote the quality of the initial solutions. A discoverer update mechanism based on sine cosine optimization and inertia weight was designed to improve the global search ability of discoverers. A followers update mechanism based on Cauchy chaos mutation was proposed to avoid the local optimum combined with chaotic mapping and Cauchy mutation. Cross Entropy minimum was used as criterion to evaluate the individual fitness. The improved sparrow search algorithm was used to find the optimal thresholds of image segmentation and realize the image segmentation. Results of image segmentation experiments show that the improved algorithm has good performance on image segmentation index and can effectively improve accuracy of image segmentation and segmentation efficiency.