Research On Road Line Recognition Algorithm Based On Machine Vision

Since the eighties of last century,autonomous navigation based on machine vision has become the main direction of intelligent vehicle driving technology research.Lane line recognition technology is one of the key technologies of autonomous navigation,domestic and foreign experts and scholars in this field of technology research done a lot of research,the purpose is to improve the recognition of robustness and real-time.In this paper,the problem of low recognition rate and inaccurate fitting of lane line under more complicated road conditions is studied.According to the region of ROI,image preprocessing,edge detection,recognition and tracking in the context of ensuring the real-time performance of the whole system,The following studies are conducted to improve the robustness of recognition.This paper first introduces the background and significance of the research,and analyzes the research at home and abroad to make clear their problems.Secondly,the road image is grayscale,and then the initial region of interest is segmented by the vertical gray mean distribution.The image smoothing,enhancement and edge detection binarization are introduced respectively in the pretreatment stage.The pavement grayscale image processing under the strong and weak illumination is researched deeply,and the edge detection operators are compared and then designed and improved.Otsu algorithm improves the lane line recognition.The lane line edge detection stage is a further extraction of the lane line.In view of the noise interference to the lane line edge recognition,the method of recognizing the edge angle and eliminating the edge is put forward,and the edge line.The traditional method of straight line detection and curve detection is analyzed,and the improvement of probability Hough transform and RANSAC algorithm is studied emphatically.Aiming at the requirement of model flexibility,a lane line model of straight line-parabola is put forward.And the method of model area allocation is designed to solve the problem that the position of curve road is uncertain.Then the least squares method is used to get the parameter of lane line model.Experiments show that this method is robust to the unstructured road.Finally,the initial lane-line image is used to predict the lane-line range of the next frame according to the slope and intercept of the straight line model,and Kalman filtering is used to avoid excessive noise.The simulation results show that the method is robust.