In order to reduce the measurement error caused by camera radial distortion in measurement system based on image, a new distortion calibration method is designed, resorting to characteristic concentric circles target. A character target rode on a two-dimensional fine platform is situated in a known position of the object side which is vertical to the optical axis, then the target images are collected and processed. A least square arithmetic is introduced to compute the fitting radius according to which the characteristic target is moved in a scheduled mode until the radius is maxim or in the demanded range. At the very time the center of the concentric circles is the camera distortion center, the coefficients of the distortion polynomial are gotten recurring to relation of arithmetical radius and imaging model radius of the camera. To find the distortion center coordinate rapidly, a coordinate alternating optimization arithmetic is applied. Experimental results show that the precision of the camera radial distortion center location demarcating can reach 0.6 pixel, and the repeated error of significant figures for polynomial coefficients is less than 0.02, and all can be realized with one character target. The distorted image can be corrected perfectly.