基于黑白环扇圆盘的角点自动识别与排序

Automatic Extraction and Sorting of Corners by Black and White Annular Sector Disk Calibration Patterns

采用传统棋盘格进行相机标定时,标定模板旋转角度对角点排序带来较大的影响,针对此设计了一种黑白环扇圆盘标定模板,并提出了与该标定模板相应的角点识别和排序算法。新的标定模板设计中心圆为1/4缺口的黑色圆,以此来消除标定模板旋转给角点排序带来的影响。通过模板区域分割算法实现标定模板的提取;创建圆形检测器剔除Harris算法提取的伪角点;利用向量积和角点几何特性完成角点排序。实验结果表明:该标定模板及其角点识别与排序算法可以有效剔除伪角点,角点排序具有良好的旋转不变性,能够实现自动相机标定。

In camera calibration with the traditional chessboard patterns, the corner sorting results are greatly influenced by the rotation angle of the calibration pattern. Therefore, a black and white annular sector disk calibration patternswas designed in this paper, and then the corner extracting and sorting algorithm was put forward correspondingly. In the new calibration pattern, the center circle is a black circle with a 1 / 4 notch for eliminating the influence of the rotation angle of the calibration pattern on corner sorting. Calibration pattern region is extracted by pattern region segmentation algorithm; circle detector is used for deleting the false corners extracted by Harris. The cross product is employed to realize the automatic corner sorting with corner geometrical property. Experiment results show that the corner detection results are correct and the sorting results show the invariance to the pattern rotation, which verifies that the calibration pattern and corresponding method are suitable for the automatic camera calibration.