基于混合色度采样率的全色度屏幕图像及视频编码研究

In recent years, with the emergence of some new computing styles, such as "cloud computing", "cloud rendering", "mobile computing", etc., screen content coding attracts great attention of researchers from both academia and industry and becomes a hot topic in video coding filed. Computer screen content can be classified into two types: computer-generated discontinuous tone content and camera-captured continuous tone content. The two types of content can be compressed effectively by different tailored methods. The majority of video CODEC software, hardware developed in the past several decades are chroma-subsampled (YUV420 or YUV422 format) ones. Unfortunately, chroma subsampling of screen image significantly degrades the visual quality of discontinuous tone content. Thus, it is necessary to compress screen image with YUV444 format. Completely adopting full-chroma coders(at least two coders, one for continuous tone content and the other for discontinuous tone content) not only significantly increases the cost of human resources but also requires tremendous investment. Since quite some portions of a compound image are continuous tone content, coding the entire image using two full-chroma coders is a overkill and too expensive. This project proposes a Mixed Chroma Sampling-rate (MCS) coding technique that combines one chroma sampling rate (such as full-chroma YUV444) coding with another chroma sampling rate (such as chroma-subsampled YUV420) coding at certain boundaries during the coding process. This is an optimal solution to achieve best 3-way balance between bit-rate, coding quality and coding complexity for screen image compression. By carrying out the following research: the strategy of mixing different Chroma Sampling-rate coding, full-chroma lossless video coding, computation complexity, rate control etc., a new screen coding scheme featuring high compression ratio, super high picture quality and low computation complexity is formed. At present, screen coding algorithm cannot implement real-time transmission of high definition screen image. This compression scheme including theory, algorithm and software implementation can well solve this problem.

随着"云计算"、"移动计算"等新型计算模式的出现，计算机屏幕图像编码成为视频编码领域的一个热门研究课题。屏幕图像由连续色调及非连续色调两种内容构成，分别适合采用不同的方法压缩。过去几十年视频压缩的软硬件技术开发都是针对色度子采样YUV420或YUV422视频。把屏幕图像色度子采样成YUV420或YUV422图像再编码会导致非连续色调内容的明显失真，必须对YUV444屏幕图像进行压缩。完全采用全色度编码（至少两个编码器，一个针对连续色调内容，另一个针对非连续色调内容）会带来巨大的人力、物力投入。本项目提出了混合色度采样率MCS的屏幕图像编码技术，通过不同色度采样率编码的混合策略、全色度无损编码、计算复杂度及码率控制等方面的研究，形成一套高压缩比、超高质量、极低复杂度的屏幕图像压缩的理论、算法及软件实现方案，最大限度降低开发成本，解决目前屏幕图像压缩算法难以满足高清屏幕实时传输的要求的难题。

"云计算"、"移动计算"等新型计算模式的出现，给视频编码领域带来了许多新的技术挑战，使计算机屏幕图像编码成为视频编码领域的一个热门研究课题。屏幕图像由连续色调及非连续色调两种内容构成，分别适合采用不同的方法压缩。本项目中，屏幕图像中的连续色调及非连续色调内容分别采用传统的混合编码器编码和串匹配方法编码。本项目对两种编码方法的无缝集成、高性能串匹配编码及算法的实用化等方面进行了全面、深入的研究。在国际上率先提出宏块级自适应LZ屏幕图像压缩的架构、算法、技术和实现，实验结果表明对某些典型屏幕视频序列，该方法较H.264编码PSNR的提高达20dB以上。在上述编码平台上整合其他无损编码方法如Run-Length Encoding(RLE)、Hextile及Portable Network Graphics(PNG)等，进一步提高了编码性能。在国际上率先提出基于双编码器混合色度采样率（Mixed Chroma Sampling-rate，简称为MCS技术）的屏幕图像及视频编码技术，并将算法整合到最新的视频压缩标准HEVC的编码平台中。研究结果表明在相同比特率的情况下，MCS编码比HEVC420编码取得明显更高的PSNR，某些图像PSNR的提高甚至超过40dB以上。紧跟国际研究前沿，基于HEVC编解码平台改进所提的屏幕图像及视频压缩算法。在一维串匹配编码的基础上，提出了仿二维匹配P2M（Pseudo-2D-matching）的概念，并可在水平和垂直两个方向上采用。实验结果表明，与传统的HEVC混合编码相比，P2M编码器取得了明显的BD-rate性能的提高；同时在相同甚至更低比特率的情况下，P2M编码器取得了明显更好的主观图画质量。提出了主要参考缓冲器PRB（primary reference buffer）和次要参考缓冲器SRB（secondary reference buffer）的概念，PRB中存储主要的参考图画数据，SRB中仅存储几个最近频繁用到的参考数据，串匹配操作即可在PRB中进行也可在SRB中进行。实验结果表明，与参考软件HM-16.4+ SCM-4.0相比，提出算法明显提高了有损和无损情况下的编码性能。本课题的研究提供了一种压缩效率高、适用范围广的屏幕图像及视频编码方法和算法，将有助于解决当前许多应用中的关键难题，具有重要的理论和实用价值。