Research On Computer Vision-based Traffic Sign Recognition

Traffic sign recognition(TSR)is an important part of intelligent transportation system(ITS)and has important application prospects in the fields of driving assistance system and driverless car.However,there are certain challenges in accurately recognizing traffic signs at present,and there are still some shortcomings in the existing traffic sign recognition algorithms.Based on summarizing the existing research methods,this paper improves some traditional traffic sign recognition algorithms.The main research contents of the paper includes as follows:(1)Due to the influence of the various external factors,traffic sign images vary in different aspects such as the shape,brightness,and the size.Therefore,the traffic sign images need to be pre-processed before the feature extraction.This paper studies the pre-processing methods of traffic sign images,including image graying,histogram equalization and the scale normalization.(2)This paper proposes a traffic sign recognition method based on HOG-LBP features and sparse representation classifier(HOG-LBP-SRC).In order to make up for the deficiency of single features,the more discriminative features,i.e.,the improved HOG features and the uniform pattern LBP features,are extracted.Then,the fusion feature is obtained by performing the generalized canonical correlation analysis method.Finally,the sparse representation classifier is used to classify the traffic signs.The experimental results show that the proposed method improves the traffic sign recognition efficiency,and the fusion feature has better classification effect than the single feature.In addition,based on this proposed method,this paper uses the MATLAB software to design the GUI interface of traffic sign recognition system.The GUI interface mainly includes three sub-interfaces: an image pre-processing interface,a feature extraction visualization interface,and a traffic sign recognition interface.(3)Aiming to the massive data and traffic sign recognition in the complex environment,a new traffic sign recognition method based on CNN multi-layer feature expression and extreme learning machine(ELM)is proposed.Firstly,the multi-layer features of traffic signs are extracted using convolutional neural network.Then,multi-scale pooling operation is used to combine the extracted feature vectors of each layer to form a multi-scale multi-attribute traffic sign feature vector.Finally,the extreme learning machine classifier is used to realize the classification of traffic signs.Experimental results show that the proposed method can effectively improve the accuracy and real-time performance in TSR.In addition,experiments in complex environments show that the proposed algorithm is robust in complex environments.