用于摄像机标定的棋盘图像角点检测新算法  

New corner detection algorithm of chessboard image for camera calibration

作　　者：杨幸芳[1,2] 黄玉美[1] 高峰[1] 杨新刚[1] 韩旭炤[1] 

机构地区：[1]西安理工大学,西安710048 [2]西安工程大学,西安710048

出　　处：《仪器仪表学报》2011年第5期1109-1113,共5页Chinese Journal of Scientific Instrument

基　　金：国家科技重大专项(No.2009ZX04001-065)资助项目

摘　　要：摄像机标定是从二维图像获取准确的三维信息的必要步骤,摄像机标定的精度在很大程度上取决于标定板控制点的图像定位精度,鉴于此,对棋盘格标定板图像的角点定位问题进行了研究。笔者通过对图像像素类型及其性质进行分析,提出了以小邻域环形模板上像素灰度的跳变次数为判据的棋盘角点检测新算法,该算法不但能快速检测出角点,而且还能确定角点的类型。利用此性质,剔除了容易受外界环境影响的棋盘外圈角点,即非X型角点,从而从另一个角度保证了所提取角点的精度。为了达到更高的精度,笔者以棋盘图像的高度对称性为依托,在每个初定位角点的局部邻域内,用灰度重心法对角点进行了亚像素定位。实验结果表明,提出的算法具有较高的精度,能够为高精度摄像机标定提供可靠数据。Camera calibration is a necessary step to obtain accurate three-dimensional information from two-dimensional images.The precision of camera calibration depends largely on the positioning accuracy of the image corresponding to the control points on calibration board;in view of this,the corner positioning problem of chessboard image is studied in this paper.Through analyzing image pixel types and their characteristics,the authors propose a new corner detection algorithm of chessboard image,which utilizes the change times of brightness on ring-shaped masks in a small neighborhood as the criterion.The proposed algorithm can not only detect the corners quickly but also distinguish the types of corners.According to this characteristic,the outermost chessboard corners,i.e.non X-type corners are excluded,which thereby ensures the precision of the detected corners from another aspect.In order to achieve sub-pixel precision,based on the high symmetry of chessboard image,the brightness gravity center method is used to position the corners in every local neighborhood of the initially positioned corner.Experimental results show that the proposed algorithm has high precision and can provide reliable data for high-precision camera calibration.

关 键 词：摄像机标定 棋盘图像 角点检测 亚像素定位 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]