The Block-Based Hybrid Video Codec

This chapter describes how transformation, quantization, motion estimation, and entropy coding can be integrated into a complete coding architecture. The limitations of the basic hybrid architecture are discussed and a collection of practical codec enhancements, that have been introduced into recent standards (such as H.264/AVC and HEVC), are described. These techniques serve to significantly enhance the performance of the basic video coding structure. The chapter describes intra-prediction—a means of enhancing performance through spatial prediction of blocks in each frame. It then extends conventional motion prediction methods through sub-pixel estimation and the use of multiple reference frames. It also examines the influence of variable block sizes and their exploitation in the coding process, both in terms of transforms and motion estimation. Finally it describes the important area of in-loop deblocking as an approach to reducing edge artifacts. Examples are provided throughout.