期刊名称：International Journal of Hybrid Information Technology
印刷版ISSN：1738-9968
出版年度：2013
卷号：6
期号：5
出版社：SERSC
摘要：Multilevel thresholding is one of the most important techniques for image processing and pattern recognition. The maximum entropy thresholding (MET) has been widely applied in multilevel thresholding. In this paper, a novel multilevel MET algorithm based on the hybrid of particle swarm optimization (PSO) and Genetic algorithm is presented. In standard PSO the non-oscillatory route can quickly cause a particle to stagnate and also it may prematurely converge on suboptimal solutions that are not even guaranteed local optimal solution. To overcome this problem, we used Genetic algorithm. To obtain an optimal solution in Genetic algorithm, operation such as selection, reproduction, and mutation procedures are used to generate next generations. The capability of this hybrid PSO that called HPGT is enhanced by cloning of fitter particles instead of worst particles that is determined based on their fitness values. The performance of HPGT algorithm and PSO algorithm compared. The results show the convergence of the HPGT is very good
