Camera lens radial distortion compensation by illuminating the epipolar lines with a projector

Abstract

Conventional radial distortion compensation methods use a calibration object or specific features within a scene. However, in this paper, we propose an automatic radial distortion compensation using a structured light system without a calibration object, a specific scene like a straight line, or knowledge of calibration parameters. When the projector illuminates straight lines along epipolar lines, the straight lines are not deformed, regardless of changes in the object shape and the relative position and orientation between the projector and the camera. Therefore, we can optimize the distortion parameters by minimizing the distance between the extracted lines in the image and the epipolar line. We can detect the lines with dynamic programming and the sub-pixel algorithm for more accurate results. The method computes the first-order radial distortion parameter, the center of radial distortion. We compared the distorted image with the undistorted image using the proposed method in the experiment. The RMS error of the straight line in the distorted image becomes smaller than threshold and the straight line in the distorted image looks like a straight line.