The extraction of the internal corners of planar checkerboard pattern image is of vital importance to camera calibration, for the accuracy of camera calibration depends on the accuracy of it. An effective approach is proposed to automatically extract the internal corners of the planar checkerboard calibration pattern image based on the characteristics of local intensity and the grid line architecture of the planar checkerboard pattern image. The proposed procedure consists of the detection of image corners, the recognition of the corners at the intersections of black and white squares and the recognition of the corners at the intersections of two groups of grid lines. The proposed approach obviously reduces the time cost for camera calibration, speeds up calibration process and is especially adapted for automatic calibration based on multiple images.