Lane Detection And Departure Warning Based On Machine Vision

With the rapid development of road traffic and the increasingly explosion in auto inventory, traffic safety has aroused widespread concern. Statistics show that most of traffic accidents caused by drivers drowsiness, inattentiveness. Driver assistance systems is the design used to be warning the driver when the vehicle leaving the lane or a collision happening, so it can help to avoid traffic accidents. The real-time lane detection and navigation system is also an essential module of the driver assistance and warning system. This paper is focus on lane detection and departure based on machine vision.Firstly, the thesis introduces the background and significance about the lane detection and departure system based on machine vision, then the research status of the thesis topic is reviewed. Finally, the framework of the lane detection and departure system based on machine vision is given.Secondly, the thesis describe the technology related to image pre-processing module such as image gray, image enhancement, image segmentation, image binarization technology. A set of the image preprocessing algorithm suitable for the thesis topic is designed base on the analysis of a large number of experiments(image enhancement select median filtering, image segmentation select canny algorithm).Thirdly, the assumptions commonly used in thesis topic is described and the lane model is established. Then the thesis introduces the two lane detection and recognition algorithm(CHEVP algorithm and Least Squares algorithm), what scene suit is also introduced for the two algorithm base on the analysis of their advantage and disadvantage. Finally, the paper gives the method for the selection of the algorithm parameters.Fourthly, the thesis gives the description of camera calibration and several common calibration methods. The simple Z. Zheng method is realized to calibrate our camera. Then TLC model is selected as the lane departure warning model base on the contrast of several classic lane departure warning models. Lastly, the paper gives the method for the selection of the TLC algorithm parameters.Finally, conclusions are given with recommendation for future work.