基于射影重建的线性摄像机自标定方法  

A Linear Camera Self-Calibration Technique Based on Projective Reconstruction

作　　者：李华[1] 胡占义[1] 

机构地区：[1]中国科学院自动化研究所 模式识别国家重点实验室,北京100080

出　　处：《软件学报》2002年第12期2286-2296,共11页Journal of Software

基　　金：国家自然科学基金资助项目(60033010;69975021);国家重点基础研究发展规划973资助项目(G1998030502)

摘　　要：提出了一种基于射影重建的线性摄像机自标定方法.该方法只需要摄像机作一次纯平移运动和两次带旋转的任意运动即可.该方法有3个主要特点.第1,由于该方法是一种线性方法,所以避免了大多数非线性方法的局部极小问题.第2,该方法是一种基于射影重建的自标定方法,由于在射影重建过程中利用了所有图像的信息,因此比以前的方法具有更好的鲁棒性.第3,该方法对硬件设备几乎没有特殊要求,在实际应用中易于实现,比如人手持一个摄像机摄取一些图像即可.模拟实验和真实图像实验均证明了该方法的可行性.In this paper, a linear camera self-calibration technique based on projective reconstruction is proposed. With the camera undergoing at least a pure translation and two arbitrary motions, all the five intrinsic parameters can be obtained. The novelties of the proposed method are three-fold. Firstly, it is a linear one, which does not involve any local minimum problem largely plagued non-linear calibration methods in the literature. Secondly, the proposed method is based on projective reconstruction. Its robustness is markedly increased since all the images are aligned in the projective reconstruction process. Thirdly, the proposed method does not require specialized hardware support, as a result, can be used in a wide range of applications. For example, in an extreme case, the camera can be merely controlled by a human arm. Simulations and experiments with real images demonstrate the applicability and feasibility of the proposed method.

关 键 词：射影重建 线性摄像机 自标定方法 三维重建 仿射重建 无穷远平面 图像匹配 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]