Error-resilient ROI coding using pre- and post-processing for video sequences

An error resilient video coding method is proposed. It yields the least amount of degradation, caused by data loss, in areas of a video frame which are of greater interest. This is achieved by employing a non-linear transform that duplicates macroblocks inside the region of interest (ROI). This increased redundancy for ROI yields a high quality reconstruction of ROI even with missing data. Simulations show that this approach has acceptable performance. Moreover, the method proposed can be implemented through pre- and post-processing of the video sequence, without modification to the source codecs (e.g. H.263, MPEG4, and H.264).