Abstract: Analyzes the shortcomings of the present X-corner detection algorithm for B/W squares on chessboard and proposes a new one in which a new BW corner detection operator is used to check out the gray values' distribution in 4 characteristic directions as defined. The coordinates of those corners are obtained with the precision at pixel level. The corner coordinates are revised to implement their localization at sub-pixel level in accordance to the similarity and influencing factors indicated in local windows. The algorithm is robust to the rotation and brightness transformation of images, and it has been applied to the X-corner images on real photos and, as a result, its validity and practicability are verified.