面向高清/超高清的感知视频编码及其并行技术研究

项目名称： 面向高清/超高清的感知视频编码及其并行技术研究

项目编号： No.61271190

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 无线电电子学、电信技术

项目作者： 门爱东

作者单位： 北京邮电大学

项目金额： 70万元

中文摘要： 更高临场感的高清/超高清视频是视频领域发展的必然。然而高质量、大尺寸的视频通信数据量巨大、计算复杂度高，必须研究更为高效的视频压缩编码。研究表明，传统的混合编码技术框架体系面临着较大的发展瓶颈，以提高计算复杂度来提升压缩效率的改进思路的发展空间越来越小。感知视频编码从图像内容的角度出发，基于人眼视觉理论来指导视频压缩，在保持图像主观感知效果基本不变的条件下能极大降低码率。 课题组将首先从视觉显著性区域优先编码准则下的恰可察失真模型展开研究，并基于上述研究提出面向高清/超高清的感知视频框架和感知优化算法。通过研究基于处理器-任务模型的感知编码并行调度算法和基于循环软件投机并行机制，课题组将致力推进面向高清/超高清视频的感知视频并行编解码算法，大幅提高编解码速率。本课题的研究对满足高清/超高清视频应用中提升压缩效率的需求和宽带移动环境下提升视频容错能力的需求都具有极其重要的理论和应用价值。

中文关键词： 感知视频编码；高清/超高清；恰可查失真；视觉显著性；并行化

英文摘要： Developing high definition (HD)/ultra high definition (UHD) video with higher vision telepresence is an inevitable tendency in video field. A more efficient video compression method should be studied, due to the huge data and the high computational complexity of the high quality and large size videos. Research showed that the joint video coding framework, based on prediction and transform architecture, faces limited development. There is little space to do research on achieving higher compression efficiency at the expense of increasing the computing complexity. Perceptual video coding, which conducts video compression on the basis of the image content and human visual system theory, can greatly reduce bit rate while maintaining the same effect of subjective perception. Firstly, our research group will carry out research on the Just-Noticeable -Distortion (JND) model which satisfies the codec priority of the visual salient region. Secondly, in view of the above analysis, we establish a HD video-oriented perceptual video coding framework and optimization algorithms. Throughout our study on the processor-task model based parallel perceptual video coding scheduling algorithm and the circulation software based speculative parallel mechanism, we will endeavor to improve the parallel perceptual video coding algorit

英文关键词： perceptual video coding；HD/UHD video；JND；visual sailency；parallel method