Research On Traffic Sign Detection Algorithm Based On Artificial Neural Network

With the progress of modern economy and society and the improvement and improvement of the people’s material living standards.Cars are becoming more and more popular,and people’s attention to road traffic safety continues to increase.Therefore,intelligent transportation system comes into people’s sight and plays a great role.The traffic sign detection is the core technology of the intelligent transportation system,so it has attracted the interest of experts and scholars all over the world and carried out many researches,and has been widely applied in driving assistance system,unmanned driving technology and other aspects.A traffic sign is a road facility that uses words or symbols to convey prohibitions,warnings or instructions,usually designed in a specific color and shape to maximize its distinction from the natural environment or other man-made background.Most traffic sign detection methods are based on color region segmentation and shape recognition.The key of color-based segmentation method lies in the choice of color space and adopts threshold value or color clustering to segment the image.Shape recognition methods usually use edge detection,mathematical morphology processing,morphological filtering,template matching and other techniques.In China,under the environment of large population base,increasing number of vehicles and gradually popularization of private cars,road problems and traffic safety problems occur frequently.Therefore,the research on traffic sign detection has very important theoretical and practical significance.However,in real traffic sign images,traffic signs will be disturbed by fading,partial occlusion,light intensity,tilt and shooting Angle,which increases the difficulty of traffic sign detection and leads to the decrease of detection efficiency and accuracy in the detection process.In order to solve these problems,many scholars have carried out in-depth research,but the research results have certain limitations.Based on the above problems,from the perspective of the shape and color of traffic signs,this paper studies and analyzes three kinds of traffic signs,namely,ban signs,warning signs and indication signs respectively,and proposes a traffic sign classification and detection algorithm based on color and shape features.The main work of this paper is as follows:1.The threshold of RGB color space is used to segment the traffic sign image,and then the edge information of traffic sign is obtained by edge detection algorithm and mathematical morphology.2.Based on edge information,a statistical feature algorithm of edge trend is proposed,which is combined with BP neural network to reflect the shape of traffic signs.3.This paper proposes a traffic sign classification and detection method based on deep neural network,which combines the statistical features of the edge trend of traffic signs with the color features.The simulation results of 500 actual traffic sign images show that the proposed method has a good detection effect for traffic sign images with different tilt angles and different shooting angles,and has a high timeliness,which achieves the expected effect.