Corner invariant and graph clustering based vehicle tracking algorithm

Vehicle tracking is an important topic in computer vision. With the development of Intelligent Transportation System (ITS), research of vehicle tracking has been more and more active. Most traditional vehicle tracking algorithms are based on background model, which are easily affected by light and perspective transform, and have difficulty to solve occlusion and camera motion. The proposed vehicle tracking algorithm tracks corner by invariant feature, and then the corner trajectories are grouped into vehicles by graph clustering based on common motion constraint. The experiment results show that the proposed algorithm can effectively perform vehicle tracking in bad environment.