Computationally efficient MCTF for scalable video coding

The discrete wavelet transforms (DWTs) applied temporally under motion compensation (MC) has recently become a very powerful tool in scalable video compression, especially when implemented through lifting. Due to bidirectional motion estimation in Motion Compensated Temporal Filtering (MCTF) framework the scalable encoder turns out to be very slow. Temporal filtering without motion compensation produces blur in the low frequency frames wherever there is a motion hence does not support temporal scalability. This paper proposes a technique that does temporal filtering without motion compensation to produce high frequency frame. The motion blocks are identified using the content of the high frequency frame and MCTF is applied only to those blocks. By applying motion estimation only to the selected blocks the technique significantly improves the speed of the encoder. In slow motion test video sequences the proposed technique performs well without compromising the quality of the video and in remaining classes of videos its complexity is reduced compared to the conventional MCTF.