Fuzzy entropy based segmentation of brain MR images using convergent heterogeneous particle swarm optimization

This paper suggests a novel idea of bi level medical image thresholding based on fuzzy entropy based segmentation using Convergent Heterogeneous Particle Swarm Optimization (CHPSO). The proposed algorithm bring in a new approach of searching the problem space by dividing the main swarm into four sub-swarms. Each sub-swarm particles will search for better solution separately that leads to better exploitation at the same time as they cooperate with each other sub-swarms to find the best global position. The outcome of the above mentioned cooperation is better exploration, convergence also to avoid local optimal best solution. Maximum fuzzy entropy is considered as the main objective function to find the best optimal threshold value and effectiveness of the proposed method is tested for four different MR Brain Images. The simulated experimental results of the proposed CHPSO based method are compared with the result of classical thresholding technique by Otsu and conventional PSO. The Comparative experimental results reveal that the proposed method using CHPSO shows better performances in terms of finding the optimal threshold value with less computation time and it outperforms Otsu and conventional PSO methods.