Research Of Road Arrows Recognition Method

Traffic environment is complex, which includes a lot of traffic information, such as traffic lights, traffic signs, zebra crossings and road arrows, and so on. It is a very difficult thing for the driver to always pay attention to all the information. Drivers misunderstand traffic information or can’t concentrate which can lead to traffic accidents, especially near the intersection. Changing lane often leads to traffic accidents, and the incidence is increasing year by year. Arrows on the road can help the driver better understand the road environment to prepare well in advance to change lanes, thus reducing or avoiding traffic accidents. Therefore, timely and accurately identify the arrows on the road is of great significance.In this paper, we get the main message through the front of the vehicle camera, and we propose a tree structure recognition algorithm based on Haar-Like features and cascade Adaboost classifier identifying arrows on the road. Firstly, we process images, including gray image and top view of transformation. Secondly, we identify image roughly. Since the road arrows are brighter than the surrounding road surface, In this paper, we propose progressive slice filtering method to extract white regions in the top view image. To some extent, which avoid the impact of the light conditions and the wearing out markings. Then calculating the number of white spots on the whole image. Getting to meet the specifications of the white candidate region through full graph traversal and integral calculation, reducing the impact of other white object appearing. Thirdly, we use a tree structure recognition algorithm based on Haar-Like features and cascade Adaboost classifier identifying arrows on the road. Finally, we use fusion algorithm for multi-frame fusion of the arrow, In this paper, we use a feature-based multi-target tracking methods to determine whether road arrows in the two images are the same or not, If a video has accumulated three image tracking fails, then we stop the program. After the program stopping, we output recognition results based on voting law. Thus we can improve the recognition rateIn this paper, we achieve the algorithm through the camera on the front of the vehicle for different tested scenarios and compare this method with other methods to detect the arrow. Experimental results show that this algorithm not only can accurately and quickly identify road arrows, but also can be applicable to the case of road arrows block or worn. Therefore, the method is suitable for practical.