视频会议系统中基于图像拼合的中间视合成  

Intermediate View Synthesis Based on Image Mosaic in Video Conferencing

作　　者：安平[1] 刘占伟[1] 刘苏醒[1] 张兆扬[1] 

机构地区：[1]上海大学通信与信息工程学院

出　　处：《上海大学学报（自然科学版）》2007年第4期371-376,共6页Journal of Shanghai University:Natural Science Edition

基　　金：国家自然科学基金资助项目(6067205260572127);上海市重点学科建设资助项目(T0102);上海市曙光计划资助项目(06SG4)

摘　　要：中间视合成是实现立体交互式显示的关键技术,即在观察者改变观看位置时,可以通过系统合成的虚拟视点获得环视的效果.该文提出一种基于图像拼合的中间视合成算法,利用图像插值和变形技术,生成虚拟视点图像.首先对立体图像对进行亮度均匀化处理,以消除左右图像间的平均亮度差异;然后根据左右视所含信息的不对称性,利用视差信息判断可靠区域,生成过渡中间视;最后采用图像拼合变形的方法进行中间视插值合成.实验表明,生成的虚拟中间视图像效果良好,基于对象的处理方式使得本算法具有速度上的优势,可用于头肩型视频会议场景的立体交互显示.Intermediate view synthesis is a key technique to realize interactive stereoscopic display. The viewer can obtain a ＂look around＂ perception from a synthesized virtual view image by changing the viewpoint. This paper proposes an intermediate view synthesis algorithm based on image mosaic to create the virtual view image by exploiting view interpolation and morphings. The original stereo pairs are first normalized to reduce luminance difference. Based on the asymmetry in the left and right camera views, credible areas are chosen to create the transitional view, exploiting the disparity information. Intermediate view images are created by combining view interpolation and morphing. Experimental results show that the proposed method can provide satisfactory rendering results with lower computation load based on object-based processing, which can be used in interactive stereoscopic display in video conferencing.

关 键 词：视频会议 中间视合成 图像拼合 过渡视点 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]