摘要：AbstractAnother kind of 3D estimation method of structural road surface based on traffic lane line information is proposed. Firstly, B-spline which is a semi-parametric model is selected to describe the lane line on the road. In the process of B-spline fitting, Image District Extraction is utilized to divide road pixels into small part Region of Interest (ROI), and Hough Transform and Least Squares Method are applied to fit short line in each ROI. The center point of the short line is taken as control point to generate B-spline curve. Finally, with the help of Similarity Transformation principle, point cloud is generated by comparing the road width between in image and on road. Experiments show that structural road can be described as 3D point cloud according to the information of lane line, and the height change of the frontal road surface can be detected from the 3D point cloud.
关键词：KeywordsDriver AssistanceSensingAutonomous VehiclesLane DetectionRoad 3D Estimation
