Automatic corner detection and localization for camera calibration

Automatic control points detection and localization is the precondition to realize camera calibration without manual aids. A novel corner detection and localization strategy is proposed in this paper, which can automatically detect and locate the corners in a chessboard pattern. A Hessian matrix detector with self-adaptive threshold is proposed to obtain the initial corners without supplied information of rows and columns number. Then symmetry constraint and a statistic technique are used to eliminate the blobs and resist the tanglesome background successfully. The whole pattern localization, which greatly reduces the computation load, is obtained by a homography method based on four automatically identified extremum corners. The proposed method has few restrictions on the chessboard image due to the above techniques. The comparison results with other traditional methods demonstrate the predominance of the proposed method, as a valuable tool for camera calibration.