文档介绍：摘 要
含噪数字识别技术综合了计算机视觉技术和模式识别技术，是智能控制系统的
重要组成部分，有着广泛的应用领域和美好的应用前景。本文以一个车辆牌照识别
系统为例进行了详尽的研究。对含噪数字识别的几个主要步骤：定位、图像预处理
和分割、字符识别，分别进行了算法的设计并对已有方案进行了比较。
主要研究了采用直接法，BP 人工神经网络法和支持向量机法(SVM)实现汽车牌
着识别系统实现的步骤和方法，取得了较高的识别率和可靠性。其中支持向量机法
在解决小样本、非线性等问题中表现出许多特有的优势。
在完成各主要步骤算法设计的基础上，建立了完整的车辆牌照识别系统。该系
统取得了较好的效果。
关键词：含噪数字识别、直接法、神经网络、支持向量机
ABSTRACT
The technique of digital recognition with noise synthesizes the technique of computer vision and
mode identify .It is an important part of intelligent controls system and has extensive application fields and
the fine applied foreground. This article takes a vehicle license plate recognition system for example to
dissect and carries on comparison of the existing methods or design for the main steps of digital recognition
with noise: positioning ,prepare processing, incise and character identify.
It mainly studies the steps and methods of realization of the system of the direct method, the BP
network method and the Support Vector Machine method (SVM) and gets a higher identifying rate and
reliability. And the SVM method takes special advantages if the solution of many problems such as the
small sample, nonlinear, etc．.
After the completion of the main steps, a whole vehicle license plate recognition system is built. And
it obtained better result.
Chu Ming-zhu (Control Theory and Control Engineering)
Directed by Prof. Yao Wan-ye
KEY WORDS: Digital recognition with noise, Directed method, BP network,
Support Vector Machine
1
目 录
中文摘要
英文摘要
第一章 绪论……………………………………………………………1
1．1 目前汽车牌照识别