Automatic Detection And Recognation Of Road Signs

With the development of the society and economic,the scale of cities was enlarged and quantity of the vehicles is increasing,so the city transportation jams and the environment worsening are becoming increasingly outstanding and traffic accidents have occurred occasionally. The attemperment and management of vehicle has become an important question of society. As a visual language, road traffic signs passed the information of road environment to the driver by graphic symbol and characters. Road traffic signs are important to guarantee the safety of pedestrian and traffic order. the purpose of Intelligent transportation is saving time, saving money and ensuring the safety of life. Traffic signs of identification can be viewed as an important part of intelligent transportation. With the development of science and technology, automobile are getting smarter, video guide is extremely popular. Traffic sign recognition can be applied to video guide of automobile. Traffic sign recognition has a certain theoretical significance and practical value.Road traffic sign has its own characteristics, we divide them into three parts according it's color and shape(red for prohibition, yellow for warnings, blue for indication signs). This paper is based upon it's color information and shape information. The main contents and innovation of this paper are as follows:(1) The road traffic signs are easily influenced by weather,illumination and surroundings under natural conditions. For the purpose of the successive research and image analysis,image enhancement is necessarily needed. The traditional method is based on color space conversion, but this paper bases on Gamma correction. We use one non-linear correction function. Experiment results confirm that improved color image enhancement is achieved. This algorithm is carried out in RGB color space,so it can satisfy the requirement of real-time performance.(2) The algorithm of segmentation of color image in every color space were studied in this paper. We present a new method of segmentation of color Image based on the differences of R,G,B. The differences of R,G,B has more regularity and its calculation is simple. Some improvements of CBH color space were presented. The experiment results show the validity and practicability the method which combined the two color space. (3) Shape is anther remarkable characteristic, we can classify the road traffic signs according to its color and shape. This paper studied the algorithm of shape analysis and traffic signs classification. The traditional method based on edge detection and template matching. This paper presents an new algorithm for shape analysis. Experiments prove that this method is quite simple and effective. Classification lay a good foundation for the recognition of road traffic signs.(4) In the recognized stage, this paper studied the recognition algorithm of traffic signs based on feature matching. This paper studies the algorithm of feature matching, Hu unchanged moment was introduced in this paper. We point out the deficiency of SIFT. This paper uses Hu unchanged moment to achieve better recognition.