光场相机标定中亚像素角点检测方法研究

Research on sub-pixel corner point detection method for light field camera calibration

针对光场相机标定中亚像素棋盘角点检测速度慢、检测精度不高等问题,该文提出一种基于双曲正切模型的亚像素精度棋盘角点检测方法。首先,采用灰度差分与模版相结合的筛选策略进行初始角点集的筛选;然后,应用一种区域灰度自适应角点检测算法获取候选角点集;最后,建立候选角点的双曲正切模型,对亚像素级候选角点的灰度分布进行拟合,通过对模型进行优化从而剔除伪角点并精确定位亚像素级棋盘角点。该方法在光场相机标定中达到98%的角点检测率,重投影误差低于0.83像素。实验结果表明,所提方法不仅可以快速提取更多的角点,同时能有效剔除伪角点,提高亚像素角点检测精度。

In allusion to problems of slow detection speed and low detection precision of sub pixel chessboard corner points during light field camera calibration,a sub pixel precision chessboard corner point detection method based on the hyper bolic tangent model is proposed in this paper. The selection strategy combining the grey difference with template is adopted to se lect the initial corner point set. The zone grey adaptive corner point detection algorithm is used to obtain the candidate corner point set. The hyperbolic tangent model of candidate corner points is set up to fit the grey distribution of sub pixel level candi date corner points. Fake corner points are eliminated and sub pixel chessboard corner points are accurately localized by optimiz ing the model. The method has a corner point detection rate of 98% and reprojection error of less than 0.83 pixels during light field camera calibration. The experimental results show that the method proposed in this paper can not only rapidly extract more corner points,but also effectively eliminate fake corner points and improve the detection accuracy of sub pixel corner points.