Occlusion detection and recovery in video object tracking based on adaptive Particle filters

Occlusion detection and recovery is a challenging task in robust real-time tracking of non-rigid objects. Particle filtering has proven very successful for non-linear and non-Gaussian estimation problems. The paper presents a method for occlusion detection and recovery for object tracking with adaptive particle filter. Firstly, object occlusion is detected with normalization factor. Secondly, adaptive transition function is employed to recovery from occlusion. Lastly, particle number is changed according to occlusion state. Experimental results show the presented method can detect occlusion and recover from it quickly.