【作者】 王东亚；

【导师】 张全法；

【作者基本信息】 郑州大学 ， 物理电子学， 2015， 硕士

【摘要】 随着现代化交通运输的飞速发展,智能交通系统作为一种先进的交通管理系统,其构建与完善日益重要。视频车辆监控系统是智能交通系统的重要组成部分,而作为视频车辆监控系统的核心,车辆牌照识别(License Plate Recognition,LPR)因其构成系统性能的瓶颈而成为研究的重点和热点。本论文详细阐述了车牌识别系统中的三个关键组成部分车牌定位、字符分割和字符识别的实现原理以及其涉及到的各种算法,并对字符识别算法深入研究。字符识别是在实现车牌精确定位,字符准确分割的基础上,利用模式识别的原理实现字符正确识别的过程。本论文在已有研究成果的基础上,对比分析了常用的模板匹配、神经网络等算法的优点与缺点,并以快速实现字符识别为目的,对基于字符笔画的字符识别算法进行改进,定义圆弧笔画,实现了车牌字符的快速识别。本文中车牌字符的识别分为两部分:汉字识别使用模板匹配,数字和英文字母采用基于笔画特征的分类识别法。其中,基于笔画特征的识别方法通过定义笔画,将字符的识别过程转化为笔画抽取、判断,并最终实现字符识别。实验证明,基于笔画特征的识别算法速度快,能够有效地区分相似字符,且具有良好的鲁棒性。更多还原

【Abstract】 With the rapid development of modern transportation, intelligent traffic system, as a kind of advanced traffic management system, becomes increasingly important. Video vehicle monitoring system is an important part of the intelligent transportation system, while vehicle license plate recognition, as a core of video vehicle monitoring system, has become a key point of research because it is the bottleneck of the vehicle license plate recognition system performance.This paper introduces the key parts of license plate recognition system: license plate location, character segmentation and character recognition, and does further research on character recognition. Character recognition is based on precisely license plate positioning and accurate segmentation character and its realization process is based on the principles of pattern recognition. This paper, on the basis of previous studies, analyses the advantages and disadvantages of commonly used algorithms, such as template matching, neural network etc. In order to realize character recognition, this paper proposes a fast recognition algorithm based on character stroke. This paper divides license plate character recognition into two parts: Chinese characters recognition using template matching and digital and English characters based on the stroke feature. The method of recognition based on the characteristics of stroke transforms the process of recognition of characters into stroke extraction and judgment, and finally realizes the character recognition.After testing, this method can effectively distinguish the similar character and is fast, besides, it has good robustness.更多还原