The chessboard pattern is very useful in camera calibration and image distortion correction. Usually

these points of intersection of every four adjoining square panes are considered as feature points. The locating precision of these points will decide the precision of calibration. However

in the real photos

the point of intersection will diffuse into a group of pixels instead of one pixel. This makes the detection with high precision very difficult. In this paper

the shortcoming of present algorithms is analyzed and a kind of new detecting algorithm is proposed. This algorithm considers the structural characteristic of chessboard pattern that the pixels are central symmetrical about the points of intersection and antisymmetric about the lines of intersection of every two adjoining square panes. The precise coordinate of feature point is decided by calculating the symmetry index and antisymmetry index for every pixel. The experiment shows this algorithm is robust to noise

projective and distortion with high precision. It is also convenient for automatic detection.