Video segmentation by MAP labeling of watershed segments

This paper addresses the problem of spatio-temporal segmentation of video s equences. An initial intensity segmentation method (watershed segmentation) provides a number of initial segments which are subsequently labeled, with a known number of labels, according to motion information. The label field is modeled as a Markov Random Field where the statistical spatial and temp oral interactions are expressed on the basis of the initial watershed segme nts. The labeling criterion is the maximization of the conditional a poster iori probability of the label field given the motion hypotheses, the estima te of the label field of the previous frame, and the image intensities. For the optimization, an iterative motion estimation-labeling algorithm is pro posed and experimental results are presented.