Abstract
Moving Objects Extraction in video is an essential technique for the high-end video coding such as object-based coding and/or video editing and video creating. There are some basic algorithms depending on requirement of extraction accuracy or processing volume, for example, the possibility of real-time processing. But, the optimal algorithm does not seem to be established in the present stage. In this report, the following Moving Objects Extraction approaches are investigated with some experimental results by the author et al., (1) region merging based on the motion model, (2) Watershed algorithm combined with active counter model, (3) Level Set Method.
