Multi-Threshold Infrared Image Segmentation Based on the Modified Particle Swarm Optimization Algorithm

Threshold extraction is the fundamental step in multi-threshold image segmentation. This paper has introduced particle swarm optimization algorithm (PSO) for threshold extraction. But when dealing with the peaky high dimension function of maximum entropy for multi-threshold image segmentation, the conventional PSO is apt to be trapped in local optima called premature. This can cause image segmentation failure. This paper proposes a modified particle swarm optimization method (MPSO), which improves convergence speed and search capacity and avoid the premature phenomena when used in threshold extraction. Simulation results show that the MPSO has better performance and quicker speed. The experimental results also show that with the modified PSO as a threshold extraction method, the image is segmented fairly well and the segmentation speed improves greatly.