摘要
本文要解决的问题是从一组对任意场景并已知拍摄视点的图片中获取三维形状。多相机场景重建问题较之立体匹配问题要复杂得多,主要是由于要考虑遮挡的因素。这篇文章中我们采用了能在立体匹配中产生很优秀的结果的方法即图形分割的能量最小化算法。我们首先构造了合适的能量方程,因为最小化能量方程实际上是个NP问题,所以我们构造一个基于图分割的近似算法来寻找局部最小值。实验结果表明了我们方法的有效性。
This paper address the problem of computing the 3D shape nf all arbitrary, scene from a set of images taken at known view-points.Multi-camera scene reconstruction is much more difficulty than stereo since it can not ignore the issue of visibility. In this paper,we take the method of energy minimization via graph cuts,which can get good results for stereo match,First we define the proper energy functinn,we know that minimizing energy function is a NP-hard problem exactly,so we construct an approximation algorithm based on graph cuts to find the local minimum.Expermental data demonstrates the effectiveness of our approach.
出处
《微计算机信息》
2009年第31期129-130,188,共3页
Control & Automation
基金
基金申请人:杨旭波
项目名称:基于投影设备的空间增强现实关键技术研究
基金颁发部门:国家863计划(2006AA01Z307)
关键词
三维重建
能量最小化
图分割
遮挡
3D reconstruction
energy minimization
graph cuts
occlusion
