Segmentation of Compressed Video Using Macro Block Classification and Global Motion Estimation

In this paper, the moving regions from the compressed video are segmented using various techniques such as Global motion estimation, markov random field model and macro block classification. First motion vectors are extracted from compressed video and are classified into different classes and global motion estimation was done on the motion vectors and coarse segmentation map was obtained. Then Motion vector quantization (VQ) based on similarity of local motion is used to find the likely number of moving regions. The inferred statistics are used to initialize prior probabilities for subsequent Markov Random field (MRF) classification, which produces coarse segmentation map. Finally, coarse to fine strategy is utilized to refine region boundaries.