Automatic Vehicle Classification Based On Image Recognition

With the increasing problems in traffic management , the development of computer technology has made the intelligent transportation system more and more applied,the automatic vehicle classification is an important subdivision of the intelligent transportation system.This dissertation studies the problem of automatic vehicle classification based on image recognition.This dissertation introduces firstly the development of vehicle classification system,presents some methods for automatic vehicle classification ,and analyzes the advantages and disadvantages of these methods; then introduces the current development of vehicle classification system based on image recognition.The principles,key problems and solutions of the system are dicussed,the current vehicle classification standards,segmentation of vehicle image and vehicle feature selection are also evaluated.<> is adopted in this dissertation as the vehicle classification standard.After image registration, the vehicle area is segmented by using the difference image of background and vehicle.At the first stage of vehicle classification,the vehicle shape parameters are extracted using the summation of line point, then the parameters are processed as the inputs of the BP neural network classifier.The classifier classifies vehicles by their shape. At the second stage of vehicle classification,the outpout of the classifier determines the further classification.This dissertation also gives the experimental result with a correct rate for vehicle classification of 81%,which shows the effectiveness of the system.