Research And System Design Of Vehicle Face Recognition Method

With the rapid development of technology and people's social life,vehicles,as a means of transportation,gradually play an important role in everyone's daily life.However,the traffic accidents caused by vehicles are continuous,and the criminal behaviors using vehicles are also increasing.The intelligent urban transportation system provides a more convenient and efficient solution for urban traffic management.Tracking vehicle information and owner information can effectively manage and investigate vehicle-related crimes and traffic accidents.Therefore,using the method of vehicle face recognition to achieve the purpose of vehicle model recognition provides a new idea and method for the smart transportation.This article is mainly to explore and study how to realize vehicle face recognition.First,it is to detect and locate the area of vehicle face in the car face image and intercept the image of the area of vehicle face.In order to extract the vehicle face features to focus on the vehicle face area,you can eliminate the interference of other content or background in the vehicle face image.Similar to face feature point detection,this paper respectively carries out custom key points annotation for these three kinds of vehicle face images,and it uses active appearance model for training and matching to achieve the processing of vehicle face alignment.Secondly,in order to realize vehicle face recognition,this paper separately studies two kinds of vehicle face recognition methods based on support vector machine and convolutional neural network to achieve the purpose of classifying vehicle type.Using the detected and aligned vehicle face image data sets,a detailed experiment was conducted using the support vector machine algorithm,and the effects of different texture features on the experimental results were tested separately.Experiments verify that the support vector machine using HOG texture features can achieve higher recognition rate.At the same time,this paper also proposes a vehicle face recognition method based on convolutional neural network,and paper introduces the structure of convolutional neural network in detail.The aligned and preprocessed vehicle face image is sent to the network for iterative training,and the experimental results of the above two vehicle face recognition methods are compared and analyzed.It can be concluded that the vehicle face recognition method based on the convolutional neural network has a higher accuracy,and the front face and the rear face have richer feature expression than the side face of the vehicle.These can more accurately distinguish the vehicle type.Finally,according to the vehicle face recognition method based on convolutional neural network in this paper,a simple module frame design is carried out for the establishment of the vehicle face recognition system.