A novel object tracking method based on a mixture model

Object tracking has been applied in many fields such as intelligent surveillance and computer vision. Although much progress has been made, there are still many puzzles which pose a huge challenge to object tracking. Currently, the problems are mainly caused by occlusion, similar object appearance and background clutters. A novel method based on a mixture model was proposed for solving these issues. The mixture model was integrated into a Bayes framework with the combination of locally dense contexts feature and the fundamental image information (i.e. the relationship between the object and its surrounding regions). This is because that the tracking problem can be seen as a prediction question, which can be solved using the Bayes method. In addition, both scale variations and templet updating are considered to assure the effectiveness of the proposed algorithm. Furthermore, the Fourier Transform (FT) is used when solving the Bayes equation to make the algorithm run in a real-time system. Therefore, the MMOT (Mixture model for object tracking) can run faster and perform better than existing algorithms on some challenging images sequences in terms of accuracy, quickness and robustness.