During the camera calibration process, it is necessary to deal with the missed detection and redundancy of checkerboard corners caused by poor illumination conditions and lens distortion. In this study, we analyze the gray distribution properties of the corners and propose an algorithm for corner detection based on the checkerboard corners' gray distribution features. To ensure that the checkerboard corners are not missed under poor illumination conditions and lens distortion, the proposed algorithm firstly extracts candidate corners using the gray distribution characteristics of the corners. Then, we improve the accuracy of the candidate corners through iteration, and eliminate the fake corners based on the gray distribution characteristics of the checkerboard corners, avoiding the redundant corners. Finally, we extract the corner coordinates of the checkerboard by combining the nearest neighbor points. Experimental results show that there is no omission and redundancy in the corners under poor illumination conditions and lens distortion. By applying the checkerboard corners extracted by the proposed algorithm in camera calibration, a mean square error of the re-projection error less than 0.1 pixel has been achieved, which is better than those provided by existing algorithms.