Improved Shi-Tomasi sub-pixel corner detection based on super-wide field of view infrared images.

In this work, we propose a new, to the best of our knowledge, corner point detection method for a super-wide field of view infrared imaging system. First, the edge of the checkerboard calibration board is detected at the pixel level by morphological operation. Second, the interpolation technique is used to refine the edge so that the edge has sub-pixel accuracy. We obtain the four checkerboard unit corners near the real corner point and average the coordinates of the four corners to indirectly obtain the coordinates of the real corner point. Meanwhile, we take pictures of the same calibration board at different angles for repeatability verification. It is proved that the improvement of our algorithm for the detection of corners of super-wide field of view infrared images is more feasible compared to the traditional algorithms.