Information-theoretic research of video coding: theory and algorithms

视频通信是当今和未来工程领域的一个主要增长领域。只要看看互联网电视、互联网视频会议和移动的视频的日益普及就知道了。然而，视频通信中的一个关键问题是视频数据量的不断增长与带宽日益短缺之间的矛盾。这在无线视频通信中尤其如此。事实上，随着无线视频的日益普及以及高清电视和其他新兴视频应用的激增，带宽短缺变得越来越严重，以至于它被视为值得我们国家关注的问题。因此，强烈需要更好的视频编码技术，其可以显著提高压缩效率，同时提供其他绿色特征，例如低复杂度（因此低编码和解码功耗）和可伸缩性。 现有的视频编码标准都是基于预测DCT的视频编码框架。这些标准的压缩性能优于它们的前辈，主要来自于它们的信号处理组件-预测/补偿和重构滤波的改进。虽然不能排除从该方向进一步增益的可能性，但是可以肯定的是，这种增益（如果有的话）对于给定的帧大小正在减小，并且将以极高的计算复杂度出现。因此，探索新的方向并开发改变未来几代视频编码的方法势在必行。 基于我们早期在无损和有损压缩以及开发图像和视频编码的信息论（IT）思想和算法方面的成功，在本研究中，我们将从IT角度系统地研究和探索改变游戏规则的视频编码方法，提出新的先进视频编码框架，开发各自的信息论，并设计新的高效视频编码算法。

Video communications is a major area of growth in engineering today and in the future. Just look at the growing popularity of Internet TV, Internet video conferencing, and mobile video. A key problem in video communications, however, is the conflict between the growing data volume of video and the increasing shortage of bandwidth. This is particularly true in wireless video communications. Indeed, with the growing popularity of wireless video and the proliferation of high definition TV and other emerging new video applications, the bandwidth shortage becomes increasingly severe to the extent that it is regarded as an issue meriting our national attention. As such, there is strong demand for better video coding technologies which can improve compression efficiency significantly while providing other green features such as low complexity (hence low encoding and decoding power consumption) and scalability. Existing video coding standards are all based on the framework of predictive DCT-based video coding. Compression performance gains of these standards over their predecessors come largely from the improvements in their signal processing components---prediction/compensation and reconstruction filtering. Although one cannot rule out the possibility of further gains from this direction, it is certain that such gains, if any, are diminishing for given frame sizes and would come at extreme high computation complexity. Therefore, it is imperative to explore new directions and develop game-changing approaches to future generations of video coding. Building on our early success in both lossless and lossy compression and in developing information theoretic (IT) ideas and algorithms for image and video coding, in this research, we will investigate and explore game-changing approaches to video coding systematically from the IT point of view by proposing new advanced video coding frameworks, developing their respective information theory, and designing new efficient video coding algorithms.