Pedestrian Tracking Using Particle Filter Algorithm
XU Fen, Gao Ming · 2010
Pedestrian tracking is a difficult task due to the complexity of environment and the irregular motion of human body. Particle Filters are advantageous on solving nonlinear problems with non-gaussian system noise. By extracting the target color-histogram features and calculating the similarity between particle candidates and target template region through discrete Bhattacharyya Coefficient, this paper presents a particle filter algorithm for pedestrian tracking. Experimental results show that the proposed algorithm outperforms Kalman tracking in almost all situations, especially when the target is occluded by other objects.