Research On Vehicle Feature Extraction Method Based On Vehicle Frontal Image

Intelligent transportation system is an important part of the traffic,and the automatic identification of the models is an important research direction in the field of intelligent transportation.As an essential part of modern transportation,the highway has been cheating in the process of charging.Based on this problem,this paper presents an algorithm for automatic recognition of vehicle face images.This paper consists of three main parts: image preprocessing,vehicle frontal feature extraction and vehicle identification.Based on the vehicle face segmentation method based on license plate location,the vehicle image is segmented by using color feature and edge detection.The vertical axis of the vehicle's face and the horizontal segmentation line and the contour tracking method are used to segment the headlights and the vehicle.For the problem of feature extraction of face image,the HU invariant moments of image and the LBP operator are used as the features of the region.In order to increase the practicability of the algorithm and reduce the complexity of the algorithm,a new algorithm of HU standard invariant moment is presented in this paper.By using the original HU invariant moment algorithm to carry out the standard processing,and then reduce the characteristic numerical interval.Compared with the typical HU invariant moment algorithm and LBP operator,the results are compared and analyzed.In the vehicle identification stage,this paper adopts the model identification algorithm based on RBF neural network.According to the characteristics of the field of vehicle identification,this paper presents a more effective RBF neural network structure,which is based on the theoretical analysis and a large number of experiments.Experiments show that the proposed method based on the standard HU invariant moment features and the RBF neural network model identification method can better detect the vehicle image,and obtain a higher recognition rate.