Image Segmentation using a Refined Comprehensive Learning Particle Swarm Optimizer for Maximum Tsallis Entropy Thresholding

Thresholding is one of the most important techniques for performing image segmentation. In this paper to compute optimum thresholds for Maximum Tsallis entropy thresholding (MTET) model, a new hybrid algorithm is proposed by integrating the Comprehensive Learning Particle Swarm Optimizer (CPSO) with the Powell’s Conjugate Gradient (PCG) method. Here the CPSO will act as the main optimizer for searching the near-optimal thresholds while the PCG method will be used to fine tune the best solutions obtained by the CPSO in every iteration. This new multilevel thresholding technique is called the refined Comprehensive Learning Particle Swarm Optimizer (RCPSO) algorithm for MTET. Experimental results over multiple images with different range of complexities validate the efficiency of the proposed technique with regard to segmentation accuracy, speed, and robustness in comparison with other techniques reported in the literature. The experimental results demonstrate that the proposed RCPSO algorithm can search for multiple thresholds which are very close to the optimal ones examined by the exhaustive search method. KeywordImage Segmentation, Maximum Tsallis entropy thresholding, Comprehensive Learning PSO, Powell’s Conjugate Gradient method