会议专题

A Boundary Refinement Technique Based on Bayesian Matting for 3DTV System

Stereo correspondence is important in 3DTV system. One of limitations of stereo is the boundary artifacts. The object boundary pixels receive colors both from the foreground and background, this leads to wrong mapping. We proposed a technique using Bayesian matting algorithm to separate the boundary pixels into foreground or background. We use the alpha value from the matting result to refine the stereo depth map. The distinguishable improvement in this paper is the localization of Bayesian matting which is used to solve the multiple-layer definition in complex depth map. Experiment result suggests that this technique reduce the boundary artifacts effectively.

3DTV boundary artifacts Bayesian matting stereo localization multiple-layer identification

Lingshan Liu Ming Xi Qingqing Yang Dongxiao Li Ming Zhang

Institute of Information and Communication Engineering, Zhejiang University, Hangzhou 310027, China Zhejiang Provincial Key Laboratory of Information Network Technology, Hangzhou 310027, China

国际会议

2010 Third Pacific-Asia Conference on Web Mining and Web-based Application(2010年第三届web挖掘和基于web应用亚太会议 WMWA 2010)

桂林

英文

396-399

2010-11-17(万方平台首次上网日期,不代表论文的发表时间)