Image Thresholding Using Mean-Shift Based Particle Swarm Optimization

In this paper, we propose a mean shift based particle swarm optimization (MS-PSO) algorithm to solve the image thresholding problem. PSO is an emerging evolutionary algorithm. However, the traditional PSO uses random number to move to the optimal position. The best position is based on random trials. Therefore, it often just detects the sub-optimal solutions due to its intrinsic stochastic behavior. The proposed MS-PSO uses mean shift procedure to obtain the more accurate position of the best solution. The experiment results show that the proposed method produces the better results than other methods.