A Human Body Part Segmentation Method Based on Markov Random Field

In order to address the problem that the human body part segmentation is vulnerable to the impact of human pose and noise for color images, this paper presents a human body part segmentation method based on Markov Random Field. In order to decrease the affect of illumination, the color space RGB of the pixels is transformed into the color space HSV. The local priori distribution of the image is described according to the equivalence of MRF and the Gibbs distribution, and the Gaussian distribution is utilized to depict the distribution of every pixel in the image, then the simulated annealing algorithm is employed to optimize the posteriori energy function to get the optimal segmentation. The experimental results show that the method can effectively extract the body parts and reduce the impact of body posture and noise on the segmentation.