Compound Markov Random Field Model Based Video Segmentation

Abstract

We present a novel approach of video segmentation using the proposed compound Markov Random Field video model. This segmentation scheme is based on the spatio-temporal approach where one MRF model is used to model the spatial image and other two MRF models take care in the temporal directions. In this modeling, edge feature in the temporal direction has been introduced to preserve the edges in the segmented images. The problem is formulated as pixel labeling problem and the pixel labels are estimated using the Maximum a Posteriori (MAP) criterion. The MAP estimates are obtained by the proposed hybrid algorithm. The performance of the proposed method is found to be better than that of JSEG method in terms of percentage of misclassification. Different examples are presented to validate the proposed approach.