Segmentation of Moving Objects with Information Feedback Between Description Levels

摘要: In real sequences, one of the factors that most negatively affects segmentation process result is existence scene noise. This impairs object which has to be corrected if we wish have some minimum guarantees success in following tracking or classification stages. this work propose a generic knowledge-based model improve process. Specifically, uses decomposition strategy description levels enable feedback information between adjacent levels. Finally, two case studies are proposed instantiate for detecting humans.