基于视频内容重组的显示适配技术研究

项目名称： 基于视频内容重组的显示适配技术研究

项目编号： No.61202320

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 计算机科学学科

项目作者： 任桐炜

作者单位： 南京大学

项目金额： 22万元

中文摘要： 随着个性化视频消费成为趋势，移动设备逐步成为观看视频的重要途径。视频显示适配技术通过自动改编视频内容来适应不同设备的显示需求。本项目紧扣视频显示适配中构建自动支持不同显示需求的视频内容组织形式这一关键科学问题，凝炼出内容相关的表示方式、用户感知驱动的改编策略和增量式的改编结果传输方案三项研究内容，重点研究基于曲面四棱柱的视频内容表示、多能量模板和质量评估指导的视频内容改编和内容与位置信息分离的增量式视频内容传输作为解决方案，并通过建设原型系统来验证研究成果，解决实用化推广中的关键技术障碍。本项目将会在面向显示适配的视频内容组织形式上取得突破，推动视频显示适配和相关研究领域的发展。本项目的研究成果可以服务于三网融合等数字内容产业需求，提高个性化视频消费的服务质量，具有很大的社会经济效益和推广应用前景。

中文关键词： 视频显示适配；显著物体检测；时空一致性；质量自动评估；

英文摘要： With the trend of personalized video consumption, mobile device has become an important approach to consume video information. Video retargeting automatically adapt video content to satisfy the display requirements of different devices. In this project, we focus on the key problem of constructing the novel video organization to automatically support different display requirements. We extract three problems from the key problem, including content-aware representation, user perception driven adaptation strategy, and incremental transmission of adaptation result. To these problems, we will research on curved quadrangular prism based video representation, multiple energy maps and quality evaluation guided video adaptation, and incremental video transmission with separated video content and location information as the solutions. We will also build a prototype system to verify the research achievements, and solve the key technical barriers in the promotion of video retargeting application. The project will acquire the breakthrough in video content organization for retargeting, and promote the development of video retargeting and other related research topics. The research achievements can serve Three-Network Convergence and other digital content industry, improve the service quality of personalized video consumption,

英文关键词： video retargeting；salient object detection；spatial-temporal coherence；automatic quality assessment；