A Robust Method for Vehicle Ego-Motion Estimation Based on Monocular Vision

A robust estimation method of ego-motion parameters is proposed for structural road.First,bilinear incremental ego-motion model is established,and the corners of lane markers and road markings are extracted as feature points.Then,the motion model is solved with the matched corners in the two sequential images,and the ego-motion parameters are estimated. The experiment results show that the presented method is less prone to the influence of illumination or backgrounds,and can estimate the parameters accurately on the structural road.Moreover,the proposed method is simple enough to meet the real-time requirements.