基于视频的微帧组合动态场景生成  

Video-based dynamic scenes generation with compound micro-frame

作　　者：余轶民[1] 许端清[1] 余轶军[1] 陈纯[1] 

机构地区：[1]浙江大学计算机科学与技术学院,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2009年第8期1372-1376,共5页Journal of Zhejiang University：Engineering Science

基　　金：国家"973"重点基础研究发展规划资助项目(2002CB312106);国家"863"高技术研究发展计划资助项目(2002AA414070)

摘　　要：针对基于视频的动态场景在时域上的不连续问题,提出了一种新的场景生成方法.以微帧为组成单元,以纹理合成特征匹配为基础,采用平滑过渡的方法达到时域的连续性,以获得具有真实感的视觉效果,同时应用无缝纹理混合方法保持视频帧的完整性和平滑性.实验结果表明,该算法具有良好的适应性,能够生成具有良好真实感的新动态场景,可以应用于三维仿真和虚拟现实等领域.A novel approach was presented for solving discontinuity of video-based dynamic scenes in temporal scope. The micro-flame was introduced as the base of texture synthesis and was taken to make up of a whole frame as a unit. Smooth transition method was applied to ensure temporal continuity for visual effect with reality. The seamless texture mixing algorithm was used to faithfully keep the integrity and smoothness of the new synthesized video frames. Experimental results were encouraging and showed that this approach is well-performed on a wide range of video sequences and can be used to make realistic scenes. This approach can also be applied to a large area, such as 3D emulation and virtual reality etc.

关 键 词：视频纹理 微帧 基于视频渲染 纹理混合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]