Research And Implementation Of Lane Recognition Algorithm Based On Edge And Positional Distribution Feature

With the development of economy, technology and the improvement of living quality, the number of vehicle is increasing, and along with it, the traffic problem becomes more obvious. To achieve traffic management, modernization management is urgently required. Intelligent traffic system is the cutting-edge research project in the area of electronic massage application field in the traffic and has important theoretical and application value. The intelligent vehicle visual navigation is an important part in intelligent traffic system, and illustrates significant research value and a good prospect of application in the aspect of preventing traffic accidents and improving active safety.Lane recognition is the most basic and important function in intelligent vehicle navigation system. This paper studies lane recognition technology based on visual images. The ideas are:First, in the state of preprocess, segmenting images by the similarity of cumulative function of image normalization histogram to improve the speed and accuracy of lane recognition, and then extracting edges to extract points on lanes’edges. Second, in the stage of lane recognition, proposing a method of extracting lanes by eliminating non lane regions. This method is based on the connectivity of road region and region growing method to extract road region, and then recognize lanes. Then using quick scan algorithm based on the relative place between lanes to extract lanes, at the same time, in the lane recognition process, we add screen method based on lanes’geometric characters to improve accuracy, and at last, calculating faraway lanes by lane prediction. Third, in the stage of lane tracking, recognizing lanes in later image by lanes’edges and color information extracted in the previous frame image to improve run efficiency and speed of algorithm.Experiments show, the algorithm which bases on lane character in this paper can detect lane region on the road well, and the detection and tracking is more accurate. At the same time, this algorithm is proved to have better robustness and efficiency.