Research On Structural Road Detection Algorithm Based On Vision

Recently, coupled with the development of the world's economy, the existing transportation system can't meet the needs. Many countries and institutions have invested so much into the intelligent transportation system(ITS).The most important technology of the ITS is the vehicle assistant driving and automatic guidance.The Lane Detection algorithm based on vision is a hotspot of the Intelligent Vehicle Navigation research. At present, although there are many vision guidance systems, none of them can meet our requirement in real-time, robustness, and practice.Firstly, the research is concentrated on the image processing methods and object recognition methods involved in the structural road detection technology. The principles and methods in low level image processing are introduced. The characteristics of these methods are compared.Secondly ,the low level processing of structural road picture is analysised. Through analysis of the camera model, a method of adverse perspective transformation is given. The effect of image perspective has been eleminated by this method. It's provide a good foundation for the road image processing. In accordance with the characteristics of the road signs and the characteristics of adverse perspective image, the methods of road signs enhancement and binarization has been figured out.Thirdly, the road model is analysed. The Hough transfer is used to imitate the road signs. The improvements of the transfer are given. The optimized binary image and probabilistic Hough transfer have been used to improve the speed.The least squares analysis has been used to improve the effect. Hough transfer and the least squares analysis have been used to improve imitate the road signs. The shackles of fixed template road could be cast off by the method. The capacity and robustness of the algorithm has been improved .At the end of this paper. The simulation environment has been established. The results of algorithm in variety of circumstances have benn analysised. The validity and reliability of the algorithm has been tested. The function of the algorithm has been analysised. And the function has been optimized so that the algorithm greatly shorten the time needed. It's provide a good foundation when the algorithm was used in embeded system.