H.264/AVC Scalable Video Extension Encoding System

• Main Authors: Tse-Min Chen, 陳澤民
• Other Authors: Ching-Lung Su
• Format: Others
• Language: en_US
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/7vs8a8

博士 === 國立雲林科技大學 === 工程科技研究所 === 103 === H.264/AVC scalable video extension (SVC) is a standard based on H.264/AVC coding technology for multiple demanded video with a single video encoder. Compared with H.264/AVC, SVC introduces scalable coding characteristics for spatial, temporal, and quality scalabilities that are suitable for unstable bandwidth communication environment or different decoding devices. However, the complexity of SVC is much higher than that of H.264/AVC. For real-time encoding by software SVC encoders, complexity reducing is necessary. In the dissertation, the low complexity Intra prediction is proposed, including fast Intra 4x4 prediction and Intra 16x16 prediction. To fast decide modes, Intra 4x4 adopts mode relationship between left, upper, and base layer (BL) macroblocks (MB). And, Intra 16x16 modes are estimated by the best mode of Intra 4x4 to speed up the Intra mode decision process. Otherwise, the hardware design of SVC encoders is the other trend for the real-time encoding purpose. Bandwidth and on-chip memory requirements are always both important issues in hardware implementations. And, the motion estimation (ME) module is critical path for two issues. The dissertation is also proposed the Inter prediction algorithm for the hardware encoder design of SVC that meets memory bandwidth and on-chip memory requirements. Clustered motion estimation and coding sequence reordering at macroblock and frame level processing are proposed and binifit to solve bandwidth and memory issues. Finally, the simulation results show the proposed Intra prediction algorithm reduces 52.53% complexity and only dropped 0.002dB video quality for Full HD resolution, under all Intra-frame coding with 3-layer (FullHD-HD-D1) spatial scalability. In addition, the proposed Inter prediction algorithm is compared to the previous research of Inter prediction, the proposed algorithm has a 49.20% lower external memory bandwidth and reduces the on-chip memory requirement by 80.45% with video quality enhancements of up to 0.087, 0.090, 0.078, and 0.070 dB for 4-layer (FullHD-HD-D1-CIF) spatial scalability, respectively.