Study On Lane Detection In Advanced Driver Assistant System

In recent years,Advanced Driver Assistance Systems,abbreviated as ADAS,have become hot topics at home and abroad.The lane detection technology,as a key technology that the system can achieve,has also received extensive attention from various countries.In this paper,the real-time tracking and detection of lane lines is taken as the research goal.Taking into account the accuracy and real-time performance of lane line detection,a lane detection algorithm based on Random Sample Consensus algorithm and an improved lane detection algorithm based on geometric features are proposed in combination with practical application scenarios.The main contributions of this article are as follows:(1)Research on Lane Detection Based on RANSAC AlgorithmThis paper proposes a lane detection algorithm based on RANSAC algorithm.The algorithm first preprocesses the image,and then uses a Random Sample Consensus Algorithm to evaluates the mathematical model of lane lines from a set of observed data including outliers in an iterative manner,and the mathematical model is used to detect lane lines.The experimental results show that this algorithm can detect lane lines more robustly in simple scenarios such as highways.(2)Research on Image BinarizationA local adaptive binarization method is proposed.Compared with the global binarization and local binarization methods,the local adaptive binarization method can retain more lane line information in lane detection.The experimental results show that the adaptive binarization method enhances the recognition effect of lane lines,reduces the missed detection rate of lane lines,and is suitable for lane line detection in complex real scenes.(3)Research on Improvement of Lane Detection Based on Geometric FeaturesThis paper proposes an improved lane detection algorithm based on geometric features.This algorithm first uses the boundary tracking algorithm to extract contours of the lane,and then adds the selection of geometric features such as area,aspect ratio,angle,and distance of lane contours,and fit the finally selected contour to a straight line to delimit the travelable area.The experimental results show that the proposed algorithm can further reduce the probability of false detection of lane lines and does not reduce the real-time performance of the algorithm.It can achieve both accuracy and real-time performance.