Abstract：Video object segmentation is the key issue in the MPEG-4 International Standard.This paper presents a new method for MPEG-4 video object segmentation from complicated scenes.In this method, the method of template matching and the object tracking method based on the motion estimation and compensation are combined.After the segmentation mask of the current frame has been generated by the motion estimation and compensation,the object colors in the initial frame are the same as the template and the object is detected in the object boundary area by template matching.So the object contour is refined.This object tracking method is able to solve the problems of occlusion and complicated scenes efficiently to certain extent and also tracks objects from an initial frame over any long time.
宋立锋;韦 岗;王群生. 基于模板匹配的视频对象分割[J]. 电子学报, 2002, 30(7): 1075-1078.
SONG Li-feng;WEI Gang;WANG Qun-sheng. Video Object Segmentation Based on Template Matching. Chinese Journal of Electronics, 2002, 30(7): 1075-1078.
2002, Vol. 30 
