基于平面模板自由拍摄的双目立体测量系统的现场标定  

Field Calibration of Binocular Stereo System Based on Planar Template and Free Snapping

作　　者：张辉[1] 张丽艳[1] 陈江[1] 赵转萍[1] 

机构地区：[1]南京航空航天大学机电学院,江苏南京210016

出　　处：《航空学报》2007年第3期695-701,共7页Acta Aeronautica et Astronautica Sinica

基　　金：国家自然科学基金(50475041);江苏省自然科学基金(BK2006192);教育部新世纪优秀人才支持计划

摘　　要：摄像机标定是基于光学摄像的立体测量技术中的一个十分重要的步骤,标定精度直接影响系统测量的精度和稳定性。本文提出并实现了一种简便易行的双目立体测量系统现场标定方法:利用一块特殊设计的具有不同大小圆形特征点的平面标定板,在无需控制任何运动参数的情况下,双目测量系统只需对标定板在不同角度自由拍摄一组图像即可方便地实现系统标定。该方法在已有单摄像机标定算法的基础上,加入对双摄像机相对位置和姿态的优化,同时考虑了镜头的非线性畸变,达到了较好的标定结果。另外,提出的图像点与其空间点的对应算法具有良好的稳定性,即使由于环境或摄像机摆放位置等因素的影响,一些标定板上的特征点不能被摄像机拍摄到,或者不能被正确识别时,仍然能进行标定工作。标定实验和系统标定后的三维重建结果验证了该方法的有效性。Camera calibration is the most important step for stereo vision measuring techniques, which effects the accuracy and stability of the measuring system directly. A practical binocular stereo calibration method is proposed in this paper. The calibration is done by just taking several images of a special-designed planar board in different orientations. Based on an existing algorithm that is mainly used for single camera calibration, the relative orientation and location between the two cameras are further optimized to adapt the algorithm to binocular stereo system. The radial distortion is also considered in the optimization. Moreover, by utilizing the circular feature points of different sizes in the calibrating board, a robust point correspondence algorithm is proposed. Even some feature points do not appear in some images or can not be recognized due to environment conditions or location change of the cameras, an accurate result could still be computed. The calibration and 3D reconstruction experiments demonstrate the validity and efficiency of the method proposed.

关 键 词：摄像机标定 双目立体测量 三维重建 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]