Investigation and Improved Configuration Settings for H.264 CGS/MGS Quality Scalable Coding

Abstract

The H.264/AVC scalable video coding (SVC) includes three kinds of video scalability, namely spatial scalability, temporal scalability, and quality scalability. Two options of quality scalability, CGS and MGS, may be used in combinations to achieve various quality layers. Using the CGS alone may incur high encoding complexity. Although the use of MGS may degrade the coding efficiency, it provides higher adaptation flexibility and reduces the coding complexity. In this paper, we investigate the CGS/MGS configurations that lead to good performance. From extensive experiments using the JSVM reference software, we suggest that MGS should be carefully employed. Although MGS always reduces the encoding complexity as compared to using CGS alone, its rate-distortion performance is unstable. Some configurations may cause an unexpected PSNR decrease with an increased bit rate. We investigate the reason and resolve the anomaly by a new CGS/MGS configuration scheme that restricts motion estimation to larger matching blocks.