Occlusion/disocclusion is one of the fundamental problems forsemantic video object (SVO) extraction because the tracking degradationfrom occlusion/disocclusion creates difficulty for the pixel-wiseaccuracy requirement and adds to the amount of user interaction requiredin off-line video editing applications. We present an approach based onthe application of a new extensive operator in a lattice of partitions,that exploits information from various views of the scene to address thedisocclusion problem by using a regional Bayesian formulation. Ourmultiview operator builds on the application of the maximum a posteriori(MAP) principle, by integrating a single view region classificationstage and a multiview stage that solves for disoccluded regions labeledas uncertain. Results on several real sequences show that this approachimproves the quality of the extracted SVOs and reduces the total amountof user interaction
展开▼
