Automatic Multilevel Thresholding Using Binary Particle Swarm Optimization for Image Segmentation

In this paper an automatic multilevel thresholding approach, based on binary particle swarm optimization, is proposed. The proposed approach automatically determines the "optimum" number of the thresholds and simultaneously searches the optimal thresholds, by optimizing a function which uses the gray level thresholds as parameters. The algorithm starts with large number initial thresholds, then, these thresholds are dynamically refined to improve the value of the objective function. The proposed method is validated by illustrative examples; comparison with the exhaustive search Otsu's and Kapur's methods shows its efficiency.