Region matching algorithm for video sequence

We propose an effective region matching algorithm for further object tracking purpose in this paper. In order to implement the matching, it is of key importance to generate natural shape regions in the video frame. The generating process can be fulfilled by two steps sequentially. Firstly, watershed segmentation algorithm is adopted based on the calculated gradient image. Then small regions are merged by our proposed spatial-temporal analysis method. The advantage of the merging is that the boundary between the moving objects and the background can be clearly divided. After doing all these work, region matching was applied according to our improved color diagram analysis approach so as to match related regions between two neighbor frames. This proposed method is especially applicable to the cases when the video's background is changeless or slightly-changed in our test. It is proved that our method is robust, and many experimental results show that satisfactory segmentation and matching effect are obtained even with some deformation in the relative regions. Since our method can provide the objects' moving information, it is very useful for performing object tracking in video sequences.