Research On Image Segmentation Based On MFO Algorithm And PCNN

PCNN has been developed on the basis of the biological visual nervous system.In image segmentation,PCNN only considers the properties of the image itself,so PCNN plays an important role in the field of image segmentation.Nevertheless,there are many parameters in the PCNN model and specifying different parameter values has a great influence on the segmentation result.In this paper,we improve the novel MFO algorithm and apply MFO and two improved MFO algorithms to optimizing the parameters of PCNN in image segmentation.Above all,MFO algorithm has better search capability,and it has the advantages of simple structure,less parameters and higher efficiency,whereas the algorithm still has the demerit of being easily trapped in local optimization.To solve this problem,two improved algorithms AMFO and ASMFO are proposed respectively in this paper.In AMFO,self-adaptive weight can be a utomatically changed so that the algorithm can get a greater search scope in the early stage and the precision of the search optimal solution can be increased in the later stage of the algorithm.In ASMFO,the simulated annealing algorithm is employed to accept new solutions with a certain probability,which can further remedy the problem that MFO is easy to fall into local optimization and will also enhance the global search ability of MFO algorithm.Then three methods are introduced to optimize the PCNN parameters for image segmentation.In optimized PCNN based on MFO,the PCNN parameters are optimized by utilizing the better search capabilities of MFO and are then passed to the PCNN model to complete the image segmentation.In optimized PCNN based on AMFO,adding self-adaptive weights based on MFO algorithm can provide better parameters for PCNN model and achieve better segmentation results.In optimized PCNN based on ASMFO,simulated annealing algorithm is adopted in AMFO,which greatly solves the local optimum of MFO and provides a guarantee for better segmentation results.Finally,a comprehensive performance study shows that AMFO and ASMFO have improved in accuracy,convergence and stability relative to MFO,the optimized PCNN methods can complete image segmentation tasks and the two methods based on improved MFOs have a better segmentation effect.