Efficient Object Tracking Based on Local Invariant Features
Yanjun Li, Jinfeng Yang, Renbiao Wu, Gong Fengxun · 2006
Object tracking has many applications in computer vision. Traditionally, to track objects successfully, motion prediction often plays an important role in tracking process. However, the cumulative error from motion prediction often leads to object losing, especially in occlusion. In this paper, an efficient method of object tracking without motion prediction is presented. Firstly, an adaptive Gaussian method is used to object detection. Then local features of the detected objects are extracted using the scale invariant feature transform (SIFT). Finally, object tracking are implemented by matching local invariant features which are learned online. The experimental results illustrate that the proposed method is capable of tracking objects under partial or severe occlusions