Based On Machine Vision Research Lane Departure Warning

Recently, with the rapid development of China’s highway and the rapid growth of vehicle ownership, traffic security issues have become increasingly prominent. And traditional traffic safety management measures are insufficient to cope with the complex driving safety problem. With the advent of driver assistance systems, it is used to solve driving safety problem in a new idea and the effective means. An important factor in traffic accidents is lane departure, which is caused by driver operation and inattention. Lane departure warnings assist system that is one of the important functions of driver assistance systems, real-time online test vehicle lane departure behavior and send a departure warning to drivers so that traffic accidents are avoided due to the lane departure. How to accurately identify the lane information and effective design the lane departure warning strategy is an important content of the lane departure warning system.In order to obtain useful information in front of vehicles, the method of image pre-processing is explored in this paper and the image processing technology is seek in an ideal environment where the weather is sunny and lane line is clear and linear. A soft of Matlab is applied to analysis and process the structured road images. Firstly the Otsu method is adopted to realize image segmentation and binary processing, which can simplify the representation of the image and then easier understand and analyze the image. Secondly median filter is used for denoising processing, to achieve the purpose of eliminating redundant information. In addition, sobel operator is used for edge detection, to highlight image information of lane edge and lane effectively. Finally, the Hough transform is used to get the lane in front of vehicle when the lane is detected. Using certain restrictions to extract the current lane, the relative parameters of lane lines in the image plane coordinate system are converted to the endpoint, slope and inclination of coordinates in the pixel plane coordinates.In this paper, the image processing method based on machine vision is proposed to identify lane information. Through summary and analysis the lane departure warning model and road imaging principle, the lane angle model is got on a two-dimensional image mapped by a three-dimensional scene. As a result of the movement of3d scene view, two-dimensional image lane Angle also changes. According to the height of viewpoints and perspectives from the lanes on the edge of the horizontal distance, the slope of the lane line is obtained in a two-dimensional image. And then the lane line parameters are refined, a lane angle model is proposed. The model can overcome the technical and structural problems, and has a certain practical value.Finally, the simulation is tested by design idea in this paper. Results showed that the lane angle will get high applicability when the angles between both sides of lane and driving direction are among the set threshold.