Abstract: Towards to the calibration problem of fisheye camera, a kind of calibration board is designed according to the camera’s wide visual field and high distortion. Then, high-precision algorithms for feature points extraction and ranking are proposed and implemented based on this kind of board. Experiments results show that the designment and algorithms proposed in this paper can not only guarantee the board’s full-visual-field imaging, but also can improve the feature’s detection precision, enhance its reliability and robustness, and provide wider angular range for image acquisition.

Key words: fisheye camera, calibration, ellipse fitting, feature points ranking

摘要： 针对鱼眼镜头的标定问题，结合其成像宽视野范围需求和高畸变特性，设计实现了适用于鱼眼标定的宽幅多特征圆标定板；基于该种标定板，给出了基于椭圆检测和参数拟合的特征点高精度检测和可靠性排序算法。实验结果表明，利用所述方法，不但能够实现标定板成像的全视野覆盖，而且能够在很大程度上提高特征点的检测精度，保证特征检测过程的可靠性和鲁棒性，同时能够提高较为宽泛的图像采集角度范围。

关键词: 鱼眼镜头, 标定, 椭圆拟合, 特征点排序