Texture-Based Adaptive Video Coding

Block-based hybrid video coding is the traditional and widespread framework in video coding standards, in which the centerpieces are predictive coding, transform coding and entropy coding. Undoubtedly, this traditional hybrid coding framework has achieved great success. However, we can still demonstrate the inevitable bottleneck in it after insightful analysis. The traditional hybrid coding framework seldom considers the factors of local characteristics and visual redundancy of pictures. The bottleneck is hindering further development of video coding technology. To break the bottleneck, texture-based adaptive video coding framework is brought forward in this paper.In this paper, some key techniques of hybrid coding framework are introduced at the beginning, they are predictive coding, transform coding and coefficient scan scheme. After the techinique review, shortcomings of hybrid coding are demonstrated in Chapter2. The shortcomings are problems can be partly solved by proposed texture-based adaptive video coding framework.In Chapter3, the concept of texture-based adaptive video coding is declaimed. And there are two important steps in the framework--texture classification and adaptive module design. Methods of texture classification in both encoder and decoder are explained latterly.Three adaptive module designs are given in Chapter4. They are texture adaptive scan for intra coding, texture adaptive1-D KL transform for intra coding and texture adaptive interpolation. Expremental results show that these three adaptive techniques do improve coding efficiency and subject quality.Texture-based adaptive video coding framework is a newborn concept. Recently, the framework is at its infancy. I hope many texture-based adaptive technologies will appear in near future.