摘要
描述了多视点视频系统中实现视点适应的算法,其目的是获得真实的三维感觉,这需要通过从立体图像对中获得虚拟视图。本文的视图生成方法是一种基于视差的图像插值技术。在视图生成过程中,为避免出现较大的空洞,首先对视差图进行预处理。然后在大基线条件下设计了一种基于可见性分析的视图生成方法。实验结果表明,利用该方法可以获得较高质量的虚拟视图。
Algorithms that are developed for multi-view video system with viewpoint adaptation are presented. The goal is to give the feeling of 3D, which is achieved by synthesis of virtual views from a stereo image pair. Our view synthesis method is an image interpolation with disparity information. In the process of view synthesis, it is necessary to pre-process the disparity map in order to avoid big hole. Then this paper design a view synthesis algorithm based on visibility analysis under large baseline condition. The experimental results show that the method can obtain virtual views with higher quality.
出处
《电子测量技术》
2009年第3期100-101,164,共3页
Electronic Measurement Technology
基金
国家自然科学基金项目(60772124)
上海大学科学技术发展基金(基于光学的动态目标三维信息获取技术研究)
关键词
多视点视频
大基线
视图生成
multi-view video
large baseline
view synthesis