A method for vehicle-mounted camera calibration under urban traffic scenes

We address the problem of vehicle-mounted camera calibration under urban traffic scenes in the fact that the traditional calibration methods are of practical restricted, since the intrinsic parameters should be calibrated in the laboratory and it is impossible for re-calibration resulting from the parameters drifting or re-focusing when driving on road. In this paper, we propose a novel method for camera parameters recovery by computing the vanishing points corresponding to the Manhattan directions, as the urban traffic scenes are usually man-made and the important lines and signs for driving are typically lying in the Manhattan directions. The lines in the scene are detected automatically and the clusters corresponding to Manhattan directions are obtained using RANSAC-like methods for general scenes. We also propose to use Bayesian inference to fit lines for the scenes where few lines lie in one Manhattan direction. Experiments are conducted in different scenes, and experimental results demonstrate the accuracy and practicability of our approach.