一种大场景下的棋盘靶标角点自动定位算法

An Automatic Location Algorithm for Chessboard Target Corners in Wide Background Environments

针对棋盘图案,提出一种亚像素精度的角点自动定位算法.该算法采用由粗到精的分层次检测策略,首先通过投影算法自动定位靶标图像中棋盘子图像的大致位置;其次在棋盘子图像中进行Hough变换,并根据角度投影图和幅值投影图进行棋盘角点的初步定位;最后在初步定位的角点邻域内通过高斯灰度插值和Harris算法得到最终的亚像素精度的角点位置.此外,本文还利用Hough变换提取的直线对角点进行自动排序.实验结果表明,该算法自动化程度较高,能够在大场景中自动定位靶标及其角点位置,而且精度较高,适用于大场景环境下畸变较小的摄像机的标定.

An automatic corner location algorithm with sub-pixel precision is proposed for chessboard pattern images. The coarse-to-fine layered detection scheme is used. Firstly, with the projection algorithm, the approximate position of chessboard sub-image in target image is located automatically. Secondly, Hough transformation is used in the chessboard sub-image, and the chessboard corner is preliminarily located according to the angle projection image and the amplitude projection image. Lastly, in the coarse corner area acquired in the second step, the final corner with sub-pixel precision is located using the Gaussian interpolation and the Harris algorithm. Furthernlore, the lines extracted with Hough transformation is used to automatically arrange the corners. Experiments show that the presented algorithm can automatically locate the chessboard target and the corners with high precision, and can be applied to the calibration of camera with less distortion in wide background environments.