Image Segmentation Based On Multi-objective Particle Swarm Optimization Algorithm

Image segmentation is a process of image preprocessing,which is the key step of image processing.The segmentation results will directly affect the later image analysis.The traditional image segmentation includes edge detection,threshold segmentation,region segmentation and clustering segmentation.According to the advantages and disadvantages of each method,we can choose different segmentation methods for different segmentation images.Therefore,there is no uniform segmentation algorithm for image segmentation.With the extensive application of swarm intelligence algorithm in image segmentation,the effect of image segmentation has been improved to some extent.However,there are still some problems,such as single segmentation result and low flexibility.In order to solve this problem,multi-objective optimization algorithm is introduced into image segmentation.Because of the poor distribution of the solution set and the poor convergence of the high dimensional target,the improved multi-objective particle swarm optimization algorithm is proposed based on the existing algorithms.First,using uniform ergodicity of Kent mapping to initialize the population,then join in the process of computing the linear regressive disturbance coefficient and chaos disturbance,finally using quick sort method to construct an improved non-dominated solution set.The time complexity of the algorithm is reduced from O(MNlogN)to O(MN).The experimental results show that the IMOPSO is superior to the classical NSGA-II.At the same time,the IMOPSO is used for image segmentation,and an image segmentation algorithm based on multi objective particle swarm optimization algorithm is proposed.IS-IMOPSO combines the FCM with K-means clustering algorithm and multi-threshold Otsu algorithm to give full play to their respective advantages.It provides a more diverse and flexible segmentation method for image segmentation.The experimental results show that the proposed method can effectively solve the problem of image segmentation and increase the flexibility and diversity of image segmentation.