The Algorithm Research Of The Automatic Vehicle Identification System

Vehicle Management System consists of three parts: vehicle sensor subsystems, automatic billing system IC card, and motor vehicle license identification system. Based on the management of the actual needs, a set of automated management systems is designed and introduces the hardware on the system design, workflow, software design and IC card management. As the car park management systems for the background, through digital image processing technology, the vehicle license automatic identification of problems are deeply studied and discussed.The image automatic vehicle identification mainly means Automatic Vehicle Identification. Because the domestic plate type is more, the environment in which vehicles stay is of different conditions, it is difficult to design a good adaptive, real-time, and the high accuracy of the vehicle license plate recognition system. The motor vehicle license identification includes three parts: image capture and access, positioning and image of the license division, license character recognition (OCR).This paper comprehensively exposits the Automatic Vehicle Identification System, including image preprocessing, the vehicle registration positioning and segmentation, character segmentation, character recognition modules, the new algorithms, new technologies and system design. In light of the situation, for some in-depth study of key technology, mainly to solve the following several aspects: (1) how quickly and accurately positioning segmentation license; (2) how precise segmentation of each character image; (3) how to design high recognition rate of recognition decisions.In regards of the license plate positioning segmentation researches, this paper raises a template matching positioning segmentation algorithm. The algorithm takes full advantage of the inherent characteristics of plate scanning images, a match with a statistical point of pixel analysis, so it can accurately and quickly detect the plates region. The characters on the license plate identification, analysis template matching OCR algorithm advantage, taking characters from Contour feature matching and neural network portfolio identification method. Against K-L transform large computing problems, the proposed improvements eigenvalues and eigenvectors calculation. According Castrillon-Candas and Amaratunga found smoother signal wavelet decomposition of sparse coefficient with the conclusions put forward by trying to select a small number of orthogonal wavelet decomposition estimates for K-L to reduce the correlation matrix dimension, the purpose of reducing computing.This paper focus on the K-L transform improved algorithm based on the license plate character recognition of the characters in the identification process, a detailed analysis of the proof conclusions. This is my main thesis of the work, through the improvement of the algorithm, to absorb the advantages of the two methods and make up for the shortfall. But for character recognition system, also introducing some very commonly used method is conductive to our license plate recognition.Finally it is discussed that ways in C++ Design and Implementation of Automatic Vehicle Identification System, and study the prospect of the work.