Object Tracking with Appearance-based Kalman Particle Filter in Presence of Occlusions

In object tracking, one of the most challenging issues is occlusion handling. Without any adaptability to this variation, the tracker may fail. To cope with it and adapt too fast, the tracking process is performed using an appearance-based tracking algorithm. And the approach, in which Kalman filtering is prepared to bring in particle filter to solve the heavy occlusion problems, can automatically select proper appearance models to track objects according to the current tracking situation. The pixel matching served as a occlusion coefficient is used in occlusion handling. These models are used to localize objects during partial occlusions, detect complete occlusions and track them robustly. The template update method is very strongly self-adaptive. The Experimental result shows that the appearance-based Kalman particle filter algorithm is able to track objects in presence of heavy occlusions satisfactorily and the computational cost is decreased.