相机标定中角点检测与定位算法研究

Research of Corners Detection and Location Algorithm for Camera Calibration

平面标定法是传统相机标定方法中较为常用的方法,标定板角点的准确检测和精确定位是平面标定法的关键;针对当前标定板角点检测算法中存在的需要人工参与、适用性弱或引入伪角点的问题,提出了基于形态学处理的角点检测和定位算法,主要通过形态学处理的方法检测棋盘的角点位置,然后经过拟合得到角点的亚像素级位置;实验验证本文算法在角点检测上具有较高的成功率,并且提高了相机标定的精度。

Camera calibration based on planar chessboard is commonly used in traditional camera calibration methods. The detection and accurate location of chessboard comers is the key of this method. Aiming to solve the problems of artificial participation, weak feasibility or importing fake comers, a novel comers detection and location algorithm based on morphological processing was introduced. Firstly, initial location of comers was found mainly by morphological processing. Then the sub-pixel location was acquired by curve fitting. Experiments shows that this method not only has a high successful rate, but also improves the accuracy of camera calibration.