基于特征点图像序列的多摄像机全局标定  

Global calibration of multiple cameras based on an image sequence with feature points

作　　者：贾倩倩[1] 王伯雄[1] 史辉[1] 罗秀芝[1] 

机构地区：[1]清华大学精密仪器与机械学系,精密测试技术及仪器国家重点实验室,北京100084

出　　处：《清华大学学报（自然科学版）》2009年第5期676-679,共4页Journal of Tsinghua University(Science and Technology)

摘　　要：多摄像机的全局标定是影响视觉测量系统精度的关键。为了降低标定过程中多次坐标变换而引起的误差,本文提出了一种基于特征点图像序列的多CCD摄像机的全局标定方法,通过时间脉冲同步,在液晶显示屏靶标上显示一系列含有特征点的图像序列的同时,多个CCD进行拍摄,对用于显示的特征图像和CCD拍摄得到的图像进行处理和计算,即可对多个摄像机进行全局标定。该方法只需一次坐标变换,克服了常规方法下特征点难以被所有摄像机全部拍到的困难,用标定结果进行三角测量,最大绝对误差为0.08mm,证明了标定方法的精确性。Global calibration of multiple CCDs is a key factor affecting the accuracy of vision measurements. This paper presents a global calibration method for multiple cameras that minimizes the error of coordinate conversion, based on image sequences with feature points. The images with feature points are shown on an LCD monitor one by one in an image sequence, with the cameras capturing each picture at the same time. The global calibration analyzes the original pictures shown on the LCD and the pictures captured by the cameras. This method needs only one coordinate conversion and overcomes the difficulty when the feature points cannot all be captured by all the cameras. Triangulation test results show that the calibration method is accurate with an error of 0. 08 mm

关 键 词：特征图像 多摄像机 全局标定 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]