A camera calibration method based on two calibration planes model is studied for the monocular vision measurement of planar object. It uses polynomials to describe the correspondence between the measurement plane coordinates and the image coordinates. The polynomial coefficients, which can be estimated by least square method, implicitly include the camera parameters and distortion parameters. A dark area problem caused by polynomial expression of the correspondence between the two planes is studied. An image coordinates translation method is proposed to solve the problem. Considering the calibration board′s thickness, the board is put on two locations which are parallel to the measurement plane with different heights, and the relationship between the two calibration planes and the image plane is calibrated respectively. Given the perpendicular distances among the measurement plane and the two calibration planes, the real object size can be computed from perspective projection principle. The experimental results verify the effectiveness of the method.