Density Propagation Based Particle Filter Algorithm for Video Object Tracking

These Video object tracking is an important topic in multimedia technologies. Particle filtering has proven very successful for non-linear and non-Gaussian estimation problems. In this paper, we proposed a novel approach for video object tracking, named by Density Propagation based Particle Filter (DP-PF). Our approach exploits color histogram to capture the features from object in the video, integrates density propagation algorithm for particle initialization together with resample technique into particle filtering and uses Bhattacharyya distance as a similarity measure for the model. Experimental results of applying the density propagation technique show improvement in tracking and robustness in recovering from partial or complete occlusions. The superior real-time performance and higher discriminative power is also demonstrated by comparison with Data-Driven Adaption based Particle Filter.