【摘要】 为了有效解决视频流媒体传输网络带宽、播出视频质量和用户实时性访问之间的矛盾,本文提出了一种基于视觉显著计算的视频流媒体渐进式表达方法.在视频内容分析和理解的基础上,首先进行场景分类和视觉敏感区域提取;然后根据编码信息确定视频序列中各帧的重要性,估计帧内片层数据重要性;最后基于视觉显著计算的结果提出一种适应网络带宽和质量可伸缩的视频流媒体渐进式表达方法.采用中粒度质量可伸缩(MGS)编码,在模拟网络测试平台上分别针对集中式和分散式视觉敏感区域视频序列进行实验研究,实验结果验证了本文提出的基于视觉显著计算的视频流媒体渐进式表达方法的正确性和有效性.更多还原

【Abstract】 In order to balance the contradiction among network bandwidth,video quality and subscriber’s real-time access,a newapproach to progressive description of video streaming based on visual saliency computation is proposed in this paper. On the basis of video content analysis and comprehension,scene classification and Visual Sensitive Region( VSR) extraction are performed firstly. Secondly,frame importance according to coding information and slice data importance are determined. Finally,based on visual saliency computation,a newapproach to progressive description of video streaming adapted to network bandwidth and quality scalability is proposed in this paper. Applying MGS Coding,experimental study is performed on video sequence with salient region and cluttered regions in network simulation platform,the experimental results showthat the progressive description based on visual saliency computation proposed in this paper is accurate and effective.更多还原