A Fast Mode Selection Algorithm for Adaptive GOP Structure in H.264 Scalable Extension

• Main Authors: Chih-Wei Chiou, 邱志偉
• Other Authors: Chia-Wen Lin
• Format: Others
• Language: en_US
• Online Access: http://ndltd.ncl.edu.tw/handle/56747754082821322055

碩士 === 國立中正大學 === 資訊工程所 === 94 === The universal multimedia access service (UMA) is an important application of internet. However, to effectively transmit video contents through networks, we must consider two issues: heterogeneous networks and various access devices. These two issues could affect the bit-rate of transmitted video streams and so the scalable video coding (SVC) is proposed for handling these two issues. Currently, for scalable video coding, H.264 scalable extension is standardized by the Joint Video Team (JVT), which is formed with ISO/IEC MPEG and ITU-T VCEG. In H.264 scalable extension, adaptive GOP structure (AGS) is used to improve coding efficiency for the video content with larger temporal variations. However, the enormous coding complexity is a major drawback of AGS. In this thesis, we propose a fast GOP mode selection algorithm to effectively reduce the coding complexity of AGS in H.264 scalable extension. AGS needs to consider motion-compensated temporal filtering (MCTF) procedures of all possible GOP sizes. Therefore, the computational complexity will enormously increase with the more of MCTF procedures. In our proposed algorithm for AGS, after executing the first MCTF procedure with a full-sized GOP (a GOP which has the maximum GOP size), we utilize two features to decrease unnecessary MCTF procedures of sub-GOPs (GOPs which have smaller GOP sizes) for reducing the coding complexity. Experimental results show that our proposed algorithm can effectively reduce the coding complexity of AGS, but with slight decrease in PSNR.