基于模板匹配的车牌识别毕业论文
2022-05-27 22:21:57
论文总字数:18871字
摘 要
在当前社会经济、科学技术迅猛发展的背景下,机动车辆的规模数量每天都在增加,而面临的交通问题则日益严峻,这便使得车牌字符自动识别技术逐渐备受人们的重视。
根据本课题的研究方向,再根据我国车牌图像固有的特点,我们选取基于模版匹配的车牌识别方法。具体步骤包括车牌图像预处理、车牌定位、车牌字符分割以及模版匹配字符识别:
- 采用二值化处理、边缘检测以及滤波处理等方法对车牌图像进行预处理;
- 对车牌定位采用统计彩色像素点多少的方法,结合实际生活中车牌的颜
色与字符等特征,采用彩色像素点统计的方法分割出合理的车牌区域,实现车牌的定位;
- 对车牌字符分割采用阈值化分割方法,选择某个位于图像灰度取值范围
内的阈值,并与图像中的像素灰度值相比较,根据两者之间的大小关系达到区域分割的目的;
- 对车牌字符识别选用模版匹配法,将处理过的车牌字符图像与模版相互
匹配比对,根据两者之间的匹配度,输出最相似的即为识别的结果。
经实验分析,可以看出本文针对车牌识别所运用的方法整体可行,有其一定的识别正确率,在执行速度方面也表现较好。综合可以得出,基于模版匹配的车牌识别技术前景广阔,有能力在未来造福人类,并产生巨大的效益。
关键词: 车牌识别 预处理 阈值化 模版匹配
Abstract
In the background of the rapid development of social economy, science and technology, motor vehicle number on the scale every day are increasing, and facing the traffic problems is increasingly serious, which make the license plate character automatic recognition technology has attracted much attention.
According to the research direction of this thesis, and according to the inherent feature of license plate images, we select the license plate recognition method based on template matching. The concrete steps include image preprocessing, license plate location, license plate character segmentation and template matching character recognition:
- Two value processing, edge detection and filtering are used to pre process the
license plate images;
- On the license location using many statistical color pixel method, combined
the characteristics of the color and character in license plate in the real life, the color pixel statistics method to segment the reasonable license plate region and realize the license plate location;
- For license plate character segmentation by thresholding segmentation
method, choose a located within the value range of image threshold, and with the image pixel gray value by comparison, according to the size relationship between the two reached the region segmentation.
- Of license plate character recognition using template matching method, the
treated license plate character image and the template are matched, and according to the matching degree between the two output most similar is recognition results.
Through the experimental analysis, it can be seen that the method of the vehicle license plate recognition is feasible, and has a certain recognition accuracy, and it also performs well in the speed of execution. It can be concluded that the license plate recognition technology based on template matching has broad prospects, and has the ability to benefit mankind in the future, and will produce the benefit of the vast number of the license plate recognition technology.
Key Words: License plate recognition; Preprocessing; Threshold; Template matching.
目 录
摘 要 I
Abstract II
目 录 III
第一章 绪论 1
1.1 研究背景 1
1.2 国内外现状 2
1.3 设计目的 3
1.4 我国车牌分析 3
1.5 文章内容与结构 4
第二章 车牌识别系统的常见方法 6
2.1 车牌定位 6
2.2 车牌字符分割 6
2.3 车牌字符识别 7
2.4 小结 8
第三章 车牌识别系统的实现 9
3.1 车牌图像预处理 9
3.1.1 图像灰度化 10
3.1.2 图像二值化 11
3.1.3 图像边缘检测 12
3.1.4 图像形态学处理 13
3.1.5 图像滤波处理 15
3.2 车牌定位 16
3.3 车牌字符分割 17
3.3.1阈值化分割方法简述 17
3.3.2 字符归一化 19
3.4 车牌字符识别 20
3.4.1 车牌字符识别简述 20
3.4.2 基于模版匹配的车牌识别简述 21
3.5 小结 22
第四章 实验分析 23
4.1车牌定位过程及分析 23
4.2车牌字符识别及分析 26
4.3车牌识别结果及分析 29
第五章 总结与展望 31
参考文献 32
附录 33
致谢 42
第一章 绪论
1.1 研究背景
在当前社会经济、科学技术迅猛发展的背景下,机动车辆的规模数量每天都在增加,而面临的交通问题则日益严峻,这便使得车牌字符自动识别技术逐渐备受人们的重视。智能交通(Intelligent Transportation System,简称ITS)系统成为广泛研究的热门领域。基于模版匹配的车辆牌照识别技术也已成为研究的热点,它在很多的领域都能被运用实现。汽车牌照作为区别机动车辆之间的唯一标识,通过智能的车牌定位及识别技术有效地识别车牌对于整个社会的自动化管理有着重要的意义。
请支付后下载全文,论文总字数:18871字