Research And Implementation Of Automatic Recognition Of Traffic Sign Based On Support Vector Machine

With the progress of technology and social development,the urban traffic congestion problem caused by popularization of cars has seriously affected the social balance.At present.it has become one of the bottlenecks of the progress of modern society,the duty to solve road congestion and ensure traffic safety is imperative under the situation.Intelligent transportation system is an integrated system engineering combines with pattern recognition,image processing and data communication technology,it can provide timely traffic security and other information.which can solve the traffic congestion and provide safe driving protection.Therefore,the automatic identification of traffic signs can improve the performance of intelligent transportation system with high scientific and practical value.The paper monitors and identifies the road traffic signs by analyzing the color and shape information of the signs.Firstly,the color image enhancement algorithm based on the luminance separation is used to enhance the image in the RGB space,and the color distortion caused by illumination is solved;Secondly,the paper realizes RGB color space preliminary segmentation through improved average algorithm,and it removes small interfering area through an area thhreshold filtering method which is a combination of experience threshold and adaptive threshold;Thirdly,it filters the possible region of traffic signs through the long and wide range of constraints to gain the region of interest,and finally to achieve automatic identification and judgment through the system.The emphasis of this paper is to realize the combination of edge direction feature values and its weights in order to get the realization of automatic shape recognition by SVM.At the same time the paper achieves the combined vector of Hu invariant moments and Zernike moments based on trying learning the Hu invariant moments and Zernike moments,and the traffic sign recognition accuracy is improved by the combined vector.It shows that weight value got by comparing the original characteristic number.the center value and the average value can improve the recognition rate of traffic signs:specific content recognition rate of the traffic signs can be improved by the combination of Zernike moment and the improved Hu moment.