The calibration of industrial cameras usually uses circular points as control points. However, there is asymmetric projection in the calibration, so that the calibration results have errors. In order to avoid the introduction of asymmetric projection error and to correct the error in an iteration form , this paper proposes a new camera calibration method by utilizing the asymmetric projection of circular centers. Firstly, the theoretical coordinate of the projection ellipse center from a circular point on the planar pattern was deduced , then the center coordinate of actual ellipse for the image of a calibration board was extracted and the projective transformation matrix was obtained by using nonlinear least square method. Finally, the camera intrinsic parameters were obtained by using all projective transformation matrixes. Experimental result reveals that the re-projection error of the calibration result of Baumer TXG12 industrial camera is just about 1/50 pixels by using proposed method. The method can complete the calibration at a time and is able to meet the precision requirement of industrial measurement in simpler calculation and higher accuracy .