摘要
引入了梯形的一个仿射不变量,并利用这个不变量,建立了梯形的相似不变量与摄像机内参数之间的约束关系.基于这个约束关系,利用摄像机内参数的知识或梯形相似不变量的知识,可以线性确定摄像机的内参数、运动参数和梯形的相似不变量.由于梯形是由一对平行线段唯一确定的,平行线段在许多场景中经常出现,因而该方法有很广泛的适用性.实验结果表明了该算法的有效性.该工作提供了一个基于平行性约束的框架,以往的基于平行四边形、平行六面体的方法都可以纳入到这个框架中.
This paper introduces an affine invariant of trapezia, and the explicit constraint equation between the intrinsic matrix of a camera and the similarity invariants of a trapezium are established using the affine invariant. By this constraint, the inner parameters, motion parameters of the cameras and the similarity invariants of trapezia can be linearly determined using some prior knowledge on the cameras or the trapezia. The proposed algorithms have wide applicability since parallel lines are not rare in many scenes. Experimental results validate the proposed approaches. This work presents a unifying framework based on the parallelism constraint, and the previous methods based on the parallelograms or the parallelepipeds can be integrated into this framework.
出处
《软件学报》
EI
CSCD
北大核心
2007年第6期1350-1360,共11页
Journal of Software
基金
国家自然科学基金Nos.60575019
60673100~~
关键词
不变量
平行性约束
摄像机标定
3D重构
invariant
parallelism constraint
camera calibration
3D reconstruction
