To reduce the negative influence of noise on the correction of camera lens distortion, a nonmetric calibration method for camera lens distortion is proposed. In this method, all curves distorted from straight lines are found out and closed by connecting their endpoints with straight-line segment, and the sum of squares of area of those closed curves is computed as distortion measurement whose minimum is determined with the aid of genetic optimistic algorithm, and thus the distortion parameters are gotten. Computation indicates that this distortion measurement offers preferable ability of anti-noise, and that genetic algorithm is able to avoid convergence to local minima. So root-mean-square (RMS) errors of rectification images are decreased. Experiment shows that RMS errors between the rectified images obtained by calibrating with each image and those obtained by calibrating with all images are small. Both computation and experiment demonstrate that this calibrating method is robust.