Vehicle tracking fusing the prior information of Kalman filter under occlusion conditions

Abstract

The occlusion in traffic scenarios have posed great challenges in vehicle tracking. Establishing effective and robust vehicle tracking algorithms under occlusion conditions is a necessary issue for many traffic applications. This paper presents a novel approach of vehicle tracking by fusing the prior information of Kalman filter. The prior information of Kalman filter is used for background update, precise morphological operation and occlusion judgment. The fusion of observation and prior information is adopted to segment vehicles under occlusion. It effectively improves the robustness of the vehicle tracking algorithm in the case of occlusion. A novel vehicle description method is proposed to express vehicle shape more accurately and thus reduces the misjudgement of occlusion. Experiments show that the proposed algorithm can accurately track vehicles even under occlusion.