适用于移动视频分享的数据压缩与质量评价新方法

项目名称： 适用于移动视频分享的数据压缩与质量评价新方法

项目编号： No.61301090

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 无线电电子学、电信技术

项目作者： 邓宸伟

作者单位： 北京理工大学

项目金额： 24万元

中文摘要： 随着移动互联网的快速发展，越来越多的人使用智能手机、平板电脑等移动设备在线分享视频资源。视频编码是移动视频分享必不可少的关键技术，也是影响用户体验的重要因素。现有的H.264,H.264 SVC，HEVC等国际标准编码复杂度高、不能实现基于视频内容和人眼视觉特性的空间分辨率与质量层可分级，因此不能很好的适用于电池容量有限、屏幕尺寸千差万别的移动终端。此外，PSNR，SSIM等评价方法不能衡量不同分辨率视频的质量，不适合评价移动视频编码的效率。针对上述问题，本课题提出一种移动视频分享新方案，设计面向任务和需求的新型编码器及感知质量评价模型。拟对该方案中面向压缩的视频信号表征、面向移动终端的低复杂度编码、基于视频内容的任意空间分辨率和质量层可伸缩编码、面向终端用户的视频体验质量评价等关键技术进行研究。与传统视频分享方案相比，该方案将更加注重用户的使用体验、更加合理利用移动设备和无线网络资源。

中文关键词： 移动视频分享；视觉信号表征；面向移动设备的数据压缩；压缩域处理；体验质量建模

英文摘要： With the rapid development of mobile Internet, more and more people use smartphones, tablets to share video content. Video coding is one of the essential key technologies for mobile video sharing, and also one of the important factors affecting the user experiences. The existing video coding standards (e.g., H.264, H.264 SVC, and HEVC) are generally of high coding complexity and can not achieve content and human visual system based spatial and quality scalability, and therefore, these codecs can not be well adapted to mobile terminals with limited power resources, and heterogenous resolutions of display. On the other hand, the existing quality metrics (e.g., PSRN,SSIM, etc.) can not measure the quality of videos with different resolutions, and thus are not applicable for the evaluation of mobile video coding efficiency.To address these issues, we propose a novel framework for mobile video sharing in this project, and some task/demand oriented new codecs and perceptual quality metrics will be developed. More specifically, the key research issues, such as compression-oriented video signal representation, mobile oriented low-complexity video coding, content-based scalable video coding for arbitrary-resolution devices, and user oriented video quality of experience (QoE) evaluation are to be investigated. Compared wi

英文关键词： Mobile video sharing；Visual signal representation；Mobile-oriented data compression；Compressed-domain processing；Quality of experience modelling