Research On Lane Detection Algorithm Of Intelligent Vehicle

Since the birth of car more than 100 years,technology innovation and cost reduction have enabled more and more families own their own cars.According to the statistics of the National Bureau of Statistics of the People's Republic of China,at of the end of 2016,the number of private cars in China was about 170 million.However,the explosive growth of private cars has brought about many problems as well as traffic congestion,traffic accidents and so on.As a new technology,intelligent driving is regarded as the key to solve these problems,and lane detection technology is a more basic part of intelligent driving technology.At present,the vision-based lane detection method is more general.Due to the complexity of the actual environment,high reliability requirements,lane detection technology seems simple,but in fact it is difficult and still full of challenges.Many vision-based lane-line detection algorithms first select an area as the region of interest(ROI),and then perform image preprocessing on this region,usually including filtering,color space transformation,perspective transformation,edge detection and other operations.Then,the lane-line detection algorithm is used to the preprocessed image,including the detection methods based on the standard Hough transform(SHT)and its derivative algorithm,as well as the methods based on the parameter fitting.These methods usually have several drawbacks,first of all,a priori regions of interest can avoid introducing much noise and get better detection effect but also inevitably reduces the general algorithm,secondly,it is difficult to be satisfied at both the detection speed and the detection result.In this paper,the image segmentation technology based on the depth fully convolution network(FCN)is introduced,which can extract the road area from the image,because the shortcoming of the traditional lane-line detection method requires to select prior ROI manually.To select the road area as a region of interest automatically instead of manually can improve the adaptability of the algorithm to different conditions.In addition,based on the traditional lane-line detection algorithm,the line detection algorithm is improved,which reduces the time complexity and improves the detection effect.The test in the end of the article shows that the method proposed in this paper can adapt to different environments and can detect lane lines quickly and efficiently.