Both maximum entropy method and Particle swarm optimization (PSO) are common threshold segmentation methods which have been used not only in image segmentation, but also in multi-threshold segmentation. Maximum entropy method is time-consuming, PSO may easily get trapped in a local optimum. In view of this concerning issue, we propose the PSO and maximum entropy are combined to make improvements on the PSO introduced in expansion model and opposition-based module. The objective functions of the maximum entropy as well as the PSO are obtained, which have improved to optimize them and search the optimal threshold combination, to achieve multi-threshold image segmentation. The results demonstrate that the new algorithm improved the segmentation speed and enhanced the robustness. And the optimizing results are stable.
Optimal Multilevel Thresholds based on Tsallis Entropy Method using Golden Ratio Particle Swarm Optimization for Improved Image Segmentation