A Feature Detection Algorithm for Autonomous Camera Calibration

This paper presents an adaptive and robust algorithm for automatic corner detection. Ordinary camera calibration methods require that a set of feature points – usually, corner points of a chessboard type of pattern – be presented to the camera in a controlled manner. On the other hand, the proposed approach automatically locates the feature points even in the presence of cluttered background, change in illumination, arbitrary poses of the pattern, etc. As the results demonstrate, the proposed technique is much more appropriate to automatic camera calibration than other existing methods.