Motion Compensated Lifting Wavelet and Its Application in Video Coding

Figure 1. 3-level MCLIFT temporal wavelet decomposition.

A motion compensated lifting (MCLIFT) framework is proposed for the 3D wavelet video coder. This is proposed in concurrent of several other researchers outside of Microsoft, with the technology named as "motion compensated temporal filtering (MCTF)". Usually, 3D wavelet video coder performs inferior to a frame prediction based video coder, such as MPEG, because motion compensation is not handled right. By using bi-directional motion compensation in each lifting step of the temporal direction, the video frames are effectively de-correlated. With proper entropy coding and bitstream packaging schemes, the MCLIFT wavelet video coder can be scalable in frame rate and quality level. Experimental results show that the MCLIFT video coder outper-forms the 3D wavelet video coder with the same entropy coding scheme by an average of 1.1-1.6dB, and outperforms MPEG-4 coder by an average of 0.9-1.4dB.

For furthur details, please refer to the related publications:

©Copyright by Jin Li, June 22, 2001