Camera calibration from road lane markings

Abstract

Three-dimensional computer vision techniques have been actively studied for the purpose of visual traffic surveillance. To determine the 3-D environment, camera calibration is a crucial step to resolve the relationship between the 3-D world coordinates and their corresponding image coordinates. A novel camera calibration using the geometry properties of road lane markings is proposed. A set of equations that computes the camera parameters from the image coordinates of the road lane markings and lane width is derived. The camera parameters include pan angle, tilt angle, swing angle, focal length, and camera distance. Our results show that the proposed method outperforms the others in terms of accuracy and noise sensitivity. The proposed method accurately determines camera parameters using the appropriate camera model and it is insensitive to perturbation of noise on the calibration pattern.