An improved algorithm for corner extraction of checkerboard in camera calibration

In automatic camera calibration with traditional chessboard patterns, the corner sorting results are usually influenced by the rotation angle of the calibration pattern. Therefore, this paper designs an improved chessboard pattern and corresponding corner extraction algorithm. In the new pattern, five nested rings are added near the four corners to determine the sorting origin.After completing the checkerboard positioning process, the improved algorithm first identifies the nested contours by five concentric circles on the four corners of the calibration plate. Then it identifies the center of each contour and determines the order of four centers. After that it uses a perspective projection to correct part of the image on the calibration plate and reorder the calibration points in order to obtain a sequence of calibration points. Experiment results show that the corner detection results are correct and the new method has a lower mean reprojection error per image.