Traffic Sign Classification Algorithm Based On Extreme Learning Machine

In recent years,serious urban traffic jams and frequent traffic accidents have seriously threatened road traffic safety in China.In order to deal with the challenges of road traffic safety,intelligent transportation systems based on computer vision have received widespread attention.Road traffic sign recognition technology is a key content in intelligent traffic system,detection and recognition of traffic signs in natural scenes achieved by computer vision and related technologies.This technology can be applied to intelligent vehicles to achieve driverless driving and plays an important role in ensuring road traffic safety.Due to the existence of color distortion,deformation,breakage and other factors in traffic signs in natural scenes,Traffic sign recognition has always been a difficult area in the field of intelligent transportation systems.Based on a large number of relevant domestic and foreign literature,this paper studies the algorithm of traffic sign recognition based on extreme learning machine.The main work is as follows:A traffic sign detection algorithm based on classification of luminance scenes and improved three-component color difference method is proposed.The method combines image scene classification,color segmentation in RGB and shape contour features.Firstly,images will be divided into four categories including backlight,dark,bright and normal scenes,according to brightness.Then gamma correction is used for image enhancement to extract the suspected targets by detection algorithm of RGB.Finally,we will select the suspected targets to find the traffic signs according to geometric features.Experimental results show that the algorithm has good real-time and robustness.A traffic sign recognition algorithm based on local features and ELM is proposed.First,the traffic signs to be identified are pre-processed using Gaussian filtering and image enhancement algorithms,and size normalized.Then,the HOG and LBP features are extracted separately to construct a joint feature vector as ELM' input training the ELM network.Finally,the trained ELM network is used for traffic sign recognition.Experimental results show that the algorithm has good real-time performance.