Research On Detection Method Of Road Marking Based On Machine Vision

With the rapid development of the economy and society,the importance of the transportation industry in the socio-economic life has been continuously highlighted,and automobiles have become a necessary tool for economic activities and people's lives.The rapid increase in the number of motor vehicles has caused congestion and frequent accidents.Safe driving assistance technology is the basis of autonomous driving technology and an important technical means to solve traffic problems.As the most important traffic sign on the road surface,the use of road marking is one of the main means of safe driving assistance technology.The technologies such as lane departure assistance and intelligent adaptive cruise in safe driving assistance technology are also based on the detection and recognition of road markings.Therefore,to continue to improve the accuracy,real-time and robustness of road marking recognition algorithms is an important issue in safe driving assistance technology and even unmanned driving applications.The main content of this article is to study the algorithm of monocular machine vision detecting and tracking road marking.The purpose is to improve the accuracy and realtime of the detection.First of all,in the image preprocessing module,the principles and advantages and disadvantages of different algorithms in each step of image preprocessing are studied and analyzed,and the edge detection algorithm is mainly studied.At the same time,according to the steps of Canny edge detection algorithm,improvements The Canny algorithm is optimized,and the method for obtaining the high and low thresholds is optimized.It does not need to be set manually.It can be automatically calculated according to the grayscale distribution of the image.From the image test results,it can improve the algorithm's environmental adaptability.Compared with the original algorithm,the detection effect is better.An optimized algorithm process under the current conditions is given: region of interest segmentation,weighted average method graying,Gaussian smoothing filter and improved Canny edge detection.This image preprocessing process effectively eliminates some noise interference for the follow-up The algorithm provides a good foundation.Then,in the pavement marking detection part,according to the shortcomings of the traditional algorithm that is easy to be misdetected under complex road conditions,based on the characteristics of the road image and the particularity of the pavement marking,on the basis of the probability Hough transform,a variety of conditions are restricted,Improved the Hough transform algorithm for detecting road markings,and tested it with the same road image data set collected locally.The results show that the improved algorithm compared with the standard algorithm of optimized parameters has improved the accuracy by 12%.Finally,in the section of pavement marking tracking,the common tracking algorithms are briefly introduced and analyzed,and the Kalman filter tracking algorithm is used to implement the tracking detection of pavement marking.By predicting and updating the parameters of the established pavement marking model,it is established The dynamic refined ROI makes the speed of the detection algorithm increase by 17.9%,and improves the real-time and robustness of the algorithm.