摘 要

Abstract 6

第1章 绪论 1

1.1本课题的研究背景 1

1.2国内外研究现状 2

1.3本文主要研究内容 3

第2章 车牌识别方法 5

2.1介绍车牌识别常用方法 5

2.1.1结构识别 5

2.1.2统计模式识别 6

2.1.3神经元法识别 7

2.2本文采用的识别方案 8

第3章 基于模板匹配的车牌识别方法设计 10

3.1图像采集 10

3.2图像预处理 10

3.3车牌定位 12

3.4倾斜校正 13

3.5字符分割 14

3.6字符识别 15

3.6.1字符识别流程概述 15

3.6.2本文字符识别流程 17

第4章 仿真试验平台介绍 20

4.1 MATLAB发展历史 20

4.2 MATLAB的语言特点 20

4.3 MATLAB仿真优点 21

第5章 车牌识别试验与分析 23

5.2模板匹配与识别 24

5.2.1模板库建立 24

5.2.2字符识别模版匹配字符识别 25

5.3 识别试验与结果分析 27

第6章总结 31

致谢 32

附录： MATLAB车牌识别主要代码： 33

参考文献 37

摘要

最近一些年，由于科技快速发展，经济形势变好，汽车变为了人们出行的重要交通工具。由于汽车数量大量提高，对于汽车的管理难度也越来越大，为了提高汽车管理的效率，智能化车牌识别十分重要，有很好的未来。因此也受到了广泛的关注。

本文在研究各种识别系统的基础上，设计了车牌识别的方案。首先要对图片进行预处理，然后基于灰度算法对车牌进行数学建模统计。通过对车牌特点的研究，本文设计了能量特征统计法对车牌进行分析，然后通过统计的特征值，与建立好的模板库进行匹配识别。但有时候，由于图像本身的特点，识别并不是完全准确，需要垂直与水平两个方位进行校正，组合匹配。文章的最后，利用数学构件MATLAB对本设计的识别算法进行软件实现，实验证明本文所设计方案识别的速度快，准确率也较主。

关键词：汽车，图像预处理，MATLAB

Abstract

With the passage of time to the 21st century, people's economic conditions are getting better and better, people are increasingly using cars to carry out the journey, make life easier, higher quality of life. But this led to the continuous increase in the number of motor vehicles, which brought a lot of traffic problems. In this one, the license plate automatic identification of the problem again and again was raised out. Through the automatic identification of the license plate, to a large extent can reduce a lot of manpower costs, the corresponding bring great value. Therefore, in recent years, the license plate of the automatic identification system more and more people become the focus of attention.

The design of the license plate on the identification of the first license plate image preprocessing, using the following methods: angle correction, binarization, edge detection, histogram transformation and other methods. A convolution energy extremum region is obtained using a license plate location method based on a gray image. This method makes full use of the contrast image of the vehicle image texture, the complex contrast, the appearance rules, etc., and then locate the license plate by selecting the connection field including the extreme line. Finally, the template matching method is used to identify the license plate character. Since some of the character shapes are close to the peak in the vertical direction, there are significant differences in the horizontal direction. Therefore, when the matching effect in the vertical direction is not conspicuous, the accuracy can be improved by matching again in the horizontal direction. This paper uses MATLAB to complete the algorithm. License plate recognition rate is higher, adaptability, real-time better.

This paper adopts the MATLAB to complete algorithm implemented. License plate recognition rate is higher, good adaptability, real-time better.