The camera lens distortion often causes image distortion during damage monitoring process of large structures, great errors will be resulted and damage detection accuracy will be reduced if the distorted image is directly used for calibration and measurement. We propose an improved non-metric distortion calibration method based on straight line characteristics to correct the distortion effectively. Firstly, we analyze the aberration model of the distorted image in actual measurement. Then, we obtain the constraint relationships with weighting factors based on the principle of linear projective invariance and three-point approximate collinearity according to the straight line features in the scene, and establish two sets of distortion correction constraint equations to solve the distortion parameters. Finally, the feasibility and validity of the proposed method is demonstrated experimentally. Root mean square error of the proposed method is improved by 0.21 pixel compared with the calibration method based on the linear feature without weighting factors.