Infrared Human Image Segmentation Using Fuzzy Havrda-Charvát Entropy and Chaos PSO Algorithm

Based on the fuzzy Havrda-Charvát entropy,a fast thresholding method was proposed to tackle the problem of human target extraction from infrared images of poor image quality.Firstly,the fuzzy Havrda-Charvát entropies of the image's background and object were defined in image fuzzy domains transformed from histogram by Z-shaped and S-shaped membership functions.Then,with combinations of the membership function's parameters as particles,a chaos particle swarm optimization(PSO) algorithm based on Tent map was presented to find the optimum threshold.The optimum threshold was calculated through the optimum combination of the fuzzy parameters,which was determined by the maximum entropy principle.Finally,the proposed method was tested on the sets of real-world infrared human images,and was compared with several classical thresholding methods.Experimental results showed the effectiveness and the robustness of the proposed method.