机器视觉标定中的亚像素中心定位算法  

Sub-pixel target location for camera calibration

作　　者：郭进[1] 刘先勇[1] 

机构地区：[1]西南科技大学信息学院,四川绵阳621010

出　　处：《传感器与微系统》2008年第2期106-108,共3页Transducer and Microsystem Technologies

基　　金：国家科技型中小企业技术创新基金资助项目(06C26225100462)

摘　　要：双目视觉检测系统中,圆形标志的中心定位精度直接影响标定的精度和整体测量的精度。传统算法定位精度较低,不能满足高精度测量的要求。采用Canny算子对标定板图像进行边缘检测,Zernike矩定位亚像素边缘,用最小二乘拟合标志中心。实验表明:该方法对圆形标志中心点提取精确且稳定,中心定位精度优于0.1个像素,系统测量平均误差小于0.02 mm,满足高精度视觉测量要求。The central accuracy of circular targets is an important factor that has direct effect on the accuracy of calibration and the accuracy of measurement in the system based on binocular stereo. The traditional operator has poor accuracy which can not reach the requirement of high-precision measurement. The edge is measured by using Canny operator, the edge is relocated with sub-pixel by means of Zernike moment operator and the targets center location using least square fitting method. The experiment proves that the center of targets can be detected with high accuracy,the central accuracy is less than 0.1 pixel,the system error is less than 0.02 mm,which can reach the requirement of measurement precision.

关 键 词：标定 亚像素 最小二乘拟合 坎尼算子 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]