Improvement Of Shuffled Frog Leaping Algorithm And Its Application To Image Segmentation

Intelligent optimization algorithm is a kind of imitation of natural intelligence.The algorithm performs well in intelligence and robustness,and has good parallelism,strong global search ability and good adaptive ability.Intelligent optimization algorithm attracts extensive attention because of its advantages.The shuffled frog leaping algorithm(SFLA)is a new intelligent optimization algorithm.This algorithm combines the advantages of the global search of particle swarm optimization algorithm and the local heuristic search of the meme algorithm.In the process of evolution,the algorithm first performs a local search,and then uses the information sharing among subpopulations to search for a global search.Combine to find out the global optimal solution.The SFLA structure is relatively simple and easy to implement,with fewer control parameters and better performance in global search capability.But the algorithm also has many problems,such as high computational complexity and inefficient optimization.In order to improve the performance of SFLA effectively,this paper studies and analyzes the algorithm,and proposes two improved schemes,and the improved algorithms are used in multthreshold image segmentation.The main contents are as follows:(1)In depth analysis of the complex update steps in SFLA sub-population,a shuffled frog leaping algorithm based on differential evolution strategy is proposed.The new metter updates all the frogs instead of only updates the worst frog every time in the subpopulation.When the subpopulation is updated,a mixed disturbance updating method based on the differential evolution strategy is used instead of the complex condition selection in the original SFLA.And introduce the example learning method to improve the optimization performance of the algorithm.In addition,the new algorithm eliminates the random update mode in the original algorithm.Experimental results show that the improved algorithm can improve the overall optimization performance.(2)In view of the shortcoming of poor sharing between the original SFLA individuals,a highly shared migration operator in the biogeography algorithm is embedded into the SFLA to form an efficient SFLA algorithm based on biological geography learning.The new improvement includes integrating the differential mutation strategy and the example learning method.Experimental results show that this improvement improves the optimization performance of SFLA.(3)Based on the analysis of the above algorithm,the application of threshold vector to solve the maximum Renyi entropy and Otsu entropy multi-threshold image segmentation in the presence of the accurate choice,the slow speed in segmentation problem,put forward two kinds of new segmentation algorithm for multi threshold image.The experimental results show that the proposed method can accurately find the optimal threshold combination of image segmentation,and the speed is faster,and the segmentation is effective.