The Multi-dimensional Double-entropy Threshold Segmentation Based On Parallel Genetic Simulated Annealing Algorithm

The threshold segmentation method has wide application in digital imaging processing,thanks to advanced characters,such as small computational amount,low complexity and high stability.With the traditional exhaustive algorithm to search for the best optimal threshold value globally,time complexity will be increasing markedly with more numbers to threshold,which limits the practical application in two-dimensional maximum entropy shareholding methods and further promotion.In order to solve the problem of time-consuming and high complexity which is arising from the application of the multi-dimensional entropy method in multi-threshold segmentation,a new algorithm based on two-dimensional entropy threshold segmentation algorithm of parallel simulated annealing genetic algorithm is proposed in this paper.With an individual fitness function of the two-dimensional entropy optimal value,parallel evolution will proceed in multiple subgroups.Besides,the two-dimensional entropy discriminant function algorithm is used as a fitness function.With the use of innovative global optimization algorithm,the threshold search purposes can be reached effectively.Parallel genetic simulated annealing algorithm is also introduced to optimize and reach maximum entropy in order to find the optimal solution threshold.Compared with traditional simulated annealing algorithm,this creative optimization algorithm can reduce the computation time by 40%-50%,which proves that our algorithm is a simple and efficient algorithm for image segmentation.According to approaches made by the proposed four simulated annealing algorithm parallelism,two efficient parallel strategies are also given analysis in this paper,including parallel-confusion relaxation parallelism and cooperative algorithm parallelism.Therefore,the parallel simulated annealing genetic algorithm can be designed with two different implementations.By a large number of experiments and simulation analysis,the effects of two innovative segmentation algorithms and real performance are compared in detail.In addition,the algorithm developed in this paper which is a type of competent efficient and practical image segmentation techniques,can ensure the accuracy of image segmentation and the optimal threshold with satisfied access speed.