Road Traffic Sign Detection And Recognition In Natural Environments

Analysis of the relevant research institutions show that about 1.3 million people around the world loss of life due to road traffic accidents,and the main causes of traffic accidents are driver drunk driving,fatigue driving,speeding and other factors related to the driver.It's shocking that traffic accidents caused huge personal property losses.And traffic safety has been widespread concern in the world.In order to effectively improve the road traffic safety and transportation efficiency,reduce the frequency of traffic accidents,intelligent transportation system(ITS)came into being.Traffic signs recognition(TSR)system,as an important branch area in ITS,has broad application prospects in unmanned vehicles,intelligent robots and driver assistance systems.Therefore,the study of traffic sign recognition system is of great academic and practical value.This paper mainly focuses on the research and discussion on the road traffic sign detection and recognition in the natural environment.The main contents of this paper include the rapid traffic sign detection based on the region of interest with MSER and SVM under the high-resolution large scene,traffic sign multi-class recognition with multi-feature fusion and traffic sign recognition system platform construction and implementation.In the respect of traffic sign detection,in order to solve the problem that the real time performance is not good when the traditional machine learning method is used to detect the traffic sign in the large scene,a fast traffic sign detection method based on MSER and SVM is proposed.Firstly,the color is enhanced according to the color information of the sign,and the sign region of interest is extracted from the enhanced image by maximally stable extremal regions(MSER)algorithm.And then the HOG feature extraction and SVM classification is only performed for the sliding window containing the region of interest of traffic sign.The method has a high improvement in the execution time,obtains higher detection accuracy,and has good robustness.In the respect of traffic sign recognition,a traffic sign recognition method with multi-feature fusion combining global features and local features is proposed to effectively improve the accuracy of the traffic sign classification.Firstly,the LBP,HOG and Gist features of the sign image are extracted,and then the method of linear combination is used to realize the feature complementary.The principal component analysis(PCA)method is used to reduce the dimension of the data.Finally,the traffic sign training and classification is carried out using support vector machine(SVM)classifier.The experimental results show that with respect to a single feature extraction classification of traffic signs,the algorithm based on multi-featured Fusion achieved higher classification accuracy,but also meet real-time requirements.Finally,we design a traffic sign recognition system to simulate the driving environment based on the wheeled robot as the main hardware,and making use of Microsoft Visual Studio 2010 and OpenCV.The system mainly includes the function modules such as image capture and processing based on monocular vision,sign detection,sign recognition and robot motion control.