A Study of Fast Mode Decision Algorithm for Motion Estimation in H.264/AVC Coding

• Main Authors: Han-Kui Chang, 張漢魁
• Other Authors: Pei-Jun Lee
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/03983337562014111854

碩士 === 國立暨南國際大學 === 電機工程學系 === 96 === H.264 is the latest video coding standard in which some new coding techniques has adopted to achive high coding efficiency. Among the techniques of H.264, Variable block-size motion estimation utilizes seven block sizes to reduce the temporal redundancy between frames. The optimal mode decision for each macroblock is made by Rate-Distortion Optimization (RDO) in H.264 inter-frame coding. However, exhaustively searching over all possible modes to choose the best coding mode for macroblock will increase the computational complexity significantly. To decrease the computational complexity and keep the coding picture quality, this thesis proposes two fast algorithms to select the coding mode such that the number of candidate modes to be searched in motion estimation is largely reduced. First one, the thesis proposes a new algorithm by fuzzy reasoning technique for inter-mode decision with two factors (motion information and edge information is generated by DCT). The other one, this thesis uses two indices, which are the correlation between macroblock mode of current frame and its corresponding macroblock mode in the previous frame, and texture characteristics with fuzzy reasoning technique to select the group of candidate mode for inter-mode decision. In the poposed algorithms, fuzzy reasoning technique is used to decrease the encoding time by the most suitable mode or group of candidate mode determination for inter-frame coding. Experimental results show that the encoding time is reduced up to 74% on method 1. Moreover, the encoding time is reduced up to 70% on method 2, and the quality is almost remained.