A 3D Lifting Based Motion Compensation Augmented Video Coder

Abstract

This study introduces a video compression method that uses spatio - temporal lifting based algorithms to compress the video signals. The temporal correlation of consecutive frames causes temporal redundancies, which are subject to lifting-like motion predictive compression. Similarly, neighboring pixels are correlated in each frame. A 3D lifting-based method that exploits both of those correlations might be used. In this study, block based motion compensatation is added to the normal 3D lifting method. Domain of motion compensation is first selected as free, and then reverse-symmetric. It is observed that reverse-symmetric motion compensation improves the performance of the 3D lifting coding.