The Study Of Vehicle Face Detection And Classification Based On Locating License Plate

Due to the presence of fake licenses,sets of licenses,unlicensed vehicles,and the requirements of special vehicles,the method of obtaining vehicle information by identifying the license plate number has its own limitations.In this paper,the method based on the license plate positioning to detect the vehicle face to identify the vehicle uses the location of the license plate to locate,does not need to identify the license plate number and it can get vehicle information in the case of without the vehicle real license plate number.The vehicle face portion of the vehicle image contains a wealth of vehicle information.Accessing vehicle face images to achieve the vehicle face detection and vehicle recognition,this method has the advantages of high efficiency and accuracy compared with the commonly used methods of image recognition based on vehicle edge or vehicle color.This paper presents the method of vehicle face detection and vehicle recognition based on license plate location.It uses the Adaboost algorithm classifier to carry out the license plate location detection and obtain the vehicle face,extract the HOG feature of the vehicle face image,and train the SVM classifier to realize the vehicle recognition.The main study contents of this paper are as follows:The Adaboost algorithm classifier is used for license plate location.The classifier gets the license plate area and the non license plate area,and completes the license plate location.After the license plate is positioned,the vehicle face area image is acquired according to the geometric model.In the process of classifier training,the image of the true and false samples is preprocessed and the image integral channel features are extracted.The feature sample is input to the Adaboost algorithm classifier,which has the license plate area recognition and positioning function,so as to construct the classifier model.For the image of the vehicle to be detected,the integrated channel feature of the image is extracted.Adaboost algorithm classifier used to detect and locate the license plate area,in accordance with the required size to obtain the standard size of the face image.It uses the sliding window method to extract the vehicle face image HOG feature,choosing a suitable square slider for the size of the vehicle face,each time the sliding window is slipped,the system extracts the HOG feature of the image block in the window,thus progressively completing the entire feature extraction process.The SVM classifier with feature dimension reduction is used to reduce the sample feature template size.Firstly,the SVM classifier is trained for a large number of local HOG features.Then,the classification results of each local classifier are grouped together into a new feature vector.The new feature vector is trained with the final SVM classifier to complete the training of the whole vehicle face identify process.The traffic image is used to simulate the simulation of the vehicle face detection and vehicle recognition process.The simulation experiment results show that this method can identify different vehicle with high detection results.And in the traffic intelligent monitoring system,public security inspection and other fields,it has a good application prospects.