A Quantisation Parameter Adaptive Mode Decision Algorithm for H.264

• Main Authors: Yu-Hung Chen, 陳毓宏
• Other Authors: Shen-Chuan Tai
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/45390492328621746046

碩士 === 國立成功大學 === 電腦與通信工程研究所 === 93 === H.264 can out-perform MPEG-4 Visual in compression efficiency but it has too higher computational complexity when compared to the older standards. In this Thesis, two efficient schemes are proposed to reduce the computation load of the H.264 encoder. The first one is called small diamond based search (SDBS), a fast block-matching algorithm, which will be applied in integer-pixel motion estimation. The SDBS algorithm consists of multiple different search strategies and could achieve stable performance over any different class of spatial detail and temporal motion videos. The other one is called quantisation parameter adaptive mode decision (QPAMD) algorithm, which is used to determine whether the prediction results of the current Inter mode is good enough to omit the investigations of the remaining Inter modes or not (in Early Termination stage), or to find the ineffective Inter predition modes and omit the investigations of them (in Mode Selection stage). The experimental results show the combination of SDBS and QPAMD could achieve a speed-up factor about 244 (for CIF format sequences) or 147 (for CCIR-601 sequences) while a slightly PSNR degradation or bitrate increase with regard to coding efficiency.