Detection And Recognition Of Road Traffic Signs

With the development of urbanization, intelligent transportation system is one of important methods to solve traffic jams. Intelligent Transportation System(ITS) is an integrated system combining the technology of communication, measurement control and computer. With the development of intelligent transportation systems, "intelligent vehicles (unmanned vehicles)"also came into being. Conversely, intelligent vehicle research also contributed to the growing maturity of the development of intelligent transport systems.Computer vision-based traffic sign recognition under the natural scene is the key to intelligent vehicle technology and one of the difficulties. Although many experts at home and abroad have spent a lot of years to study, the problem is still not a good solution for this situation. In this paper, traffic sign detection and recognition have been researched in the following aspects:(1).We collected the traffic signs of natural scenes are RGB color space image. According to the color characteristics of traffic signs, the image in RGB space was analyzed and found that the difference between 3 channels to maintain a certain range, therefore, according to this feature through the appropriate threshold to achieve good results.(2) In this paper, the first step for further analysis of the segmented image, image segmentation that occurs after a lot of noise, according to the subsequent recognition as a great obstacle. Analysis showed that the size of traffic signs in a certain range, the paper together with the region to fill and the area threshold method to remove a large area and small area of the noise points. Further observed features of traffic signs, traffic signs can easily find the shape of the geometry are the rules, for example, the ban signs, its shape is round, from a mathematical point of view, showing circular boundary of the slope of the characteristics of a certain regularity. This article features the shape of traffic signs through the proposed changes based on statistics of tangent shape of traffic signs detection algorithm to accurately locate traffic signs.(3) Traffic sign recognition will be detected to determine what kind of traffic signs for the process. Traffic sign recognition has many ways, for example: template matching algorithm, cluster analysis, artificial neural networks. The latest research at home and abroad is based on SIFT-matching recognition algorithm. SIFT-matching algorithm which can overcome the image of the rotation, size, and the effect of light is considered nowadays the most popular method. This article is based on SIFT algorithm was proposed based on A-SIFT matching algorithm for traffic sign recognition.Experimental results show that the traffic sign recognition algorithm has better robustness.