Research On Structured Road Recognition Algorithm Based On Image Processing

With the growth of car ownership,traffic accidents happen frequently,driverless driving began to appear.Road recognition is a very important part of intelligent driving.Accurately identifying the lane line is helpful to regulate the driving behavior of drivers and ensure driving safety.In this paper,lane line recognition for structured road is carried out.According to the characteristics of structured road and its complex environmental conditions,the accuracy of recognition algorithm and the adaptability to various environments are used for related research.Firstly,for the preprocessing of road images,an adaptive region of interest calculation method is proposed,which separates the road region from the non-road region,eliminates the noise interference and improves the efficiency of lane line recognition;aiming at the problem that the lane line is not obvious under different light intensities,the grayscale stretching is proposed for the road images with different light intensities,so that the transformed lane lines can be easier to recognize.Secondly,an improved Canny algorithm is proposed for the edge detection of road images.Aiming at the problems of weakening important edge and poor adaptability in traditional Canny edge detection algorithm,this paper adopts bilateral filter instead of Gaussian filter,uses multi direction gradient template to calculate the image gradient,and uses the maximum between-class variance method to confirm the high and low thresholds of the image.These improvements effectively improve the denoising effect of the algorithm and the accuracy of edge detection.Finally,a hybrid road recognition model is proposed.For the recognition of straight road,the improved cumulative probability Hough transform is used to eliminate the road noise to adapt to the complex road environment;in order to improve the fitting accuracy of the lane line,the improved least square method is used to further fit the straight road.For the recognition of curve road,the improved RANSAC algorithm is used to recognize curve road.At the same time,in order to avoid the problem of lane line detection failure caused by road interference,the Kalman filter is used to track the lane line.Through a comparative experiment of road recognition on the public data set Tusimple,the structured road recognition algorithm proposed in this paper has a significant improvement in accuracy compared with other excellent methods,and its processing speed is also more advantageous,which has certain research value and significance.