Research On Lane Line And Target Vehicle Detection Algorithm Based On Vision

With the development of intelligent transportation systems and the advancement of image processing technology,vision-based autonomous driving technology also been rapidly developed,and the development of intelligent vehicles equipped with autonomous driving technology has also received widespread attention.The lane line and target vehicle detection algorithm is one of the core technologies of intelligent vehicles,at present,it has become a research hotspot of intelligent vehicles and a variety of algorithms have been applied to intelligent vehicles.However,the actual road traffic conditions are more complicated,and the detection of lane lines and target vehicles is susceptible to factors such as the road environment,it is difficult to accurately detect lane lines and target vehicles under different lighting conditions.In view of the above problems,this thesis studies the lane line and target vehicle detection algorithm based on vision.First,use homomorphic filtering to enhance the road image to eliminate the uneven illumination problem of the image,and reduce the impact of lighting conditions on target detection;Then divide the road image to select the lane line area,and adaptively and accurately extracting the edge features of lane line by improving the double threshold selection of Canny operator;Complete straight line detection by use the Hough transform with polar angle constraint and get the lane line detection result;Finally,get the vehicle’s current available driving area information according to the lane line parameter.The moving target in front of the road is the main threats to the driving safety of vehicles,therefor,completed the target vehicle detection by segmenting the rear shadow on the basis of lane line detection.First,according to the gray level characteristics of the rear shadow,calculate the image segmentation threshold by use the Kmeans clustering algorithm combined with quadratic threshold segmentation;Then use morphological operations to optimize the shape of the shadow feature,and obtain the coordinates of the rear shadow line by calculating the smallest circumscribed rectangle of the connected domain;Finally,generate vehicle hypothesis area according to the rear shadow line coordinates,and use the symmetry of the vehicle rear to verify the vehicle hypothesis area.Select road images under different lighting conditions from actually collected and the i Roads dataset as experimental materials to verify the proposed algorithm,the experimental material contains the road image under different lighting conditions.A large number of experiments shows that the proposed algorithm can accurately detect lane line and ahead target vehicles under different road environments,can solve the unstable detection problemunder different lighting conditions,reduce the influence of different lighting conditions on the detection accuracy of target vehicles,and the proposed algorithm has good stability.