Target tracking using two-stage sparse coding

Target tracking is an important issue in computer vision for its widely applications in video surveillance, human-computer interaction, robotic navigation, image compression and so on. There exist challenging problems such as occlusion, pose change, illumination change etc. in real videos. The two-stage sparse coding method for target tracking is proposed in this paper. The proposed method is based on Bayesian network and sparse coding with dynamic dictionary. In order to overcome the visual drift, the two-stage method is applied with target and dictionary updating to realize tracking accurately and quickly. Some public and popular challenging sequences are used to demonstrate the effectiveness of the proposed method. The experiment results show that the proposed method has better performance compared with other state-of-the-art methods.