基于信息融合的视频对象分割方法综述

Video object segmentation is one of the basic components of video retrieval, video coding, video storage and transmission. Many related algorithms have been proposed in the literature. Due to the complexity involved in video processing, a universally “best” segmentation method does not exist. Most of the proposed algorithms are concerned with special applications or constraints. Video object segmentation based on information fusion takes advantage of several different but complementary kinds of useful information, e.g. temporal information, spatial information or frequential information, selects the appropriate complementary algorithms and designs rules for fusion to find a more efficient and effective way of segmentation. This paper presents a survey of the video object segmentation techniques based on information fusion. State-of-the-art research on video object segmentation based on information fusion is first presented and analyzed. Then the current trends and future direction in this track are discussed.