Performance comparison of MCTF type codecs and hybrid DCT based codecs

A conventional video-coding system codes a video sequence at a desired fixed bit rate that is adequate for a given application. Serving different clients requires transcoding of a given sequence. Scalable coding provides a straightforward solution for video coding that can serve a broad range of applications without the need for transcoding. Current international video-coding standards do not provide fully scalable methods. The latest video-coding standard, H.264/Advanced Video Coding (AVC) provides the best rate-distortion (R-D) performance. The motion compensated embedded zero block context (MC-EZBC) coder is a motion compensated temporal filtering (MCTF) type video codec that is fully scalable and provides R-D performance that is comparable to that of MPEG-4 fine granular scalability. The paper compares the performance of the MC-EZBC codec with that of the H.264/AVC video codec. It is observed that the R-D performance of H.264/AVC is superior to that of MC-EZBC for low resolution video sequences and it is comparable in the case of high-resolution sequences in terms of luminance PSNR values, but, as in other cases, the introduction of scalability may degrade the performance of H.264/AVC.