汽车牌照识别技术的研究

发布时间：2018-01-08 14:39

本文关键词：汽车牌照识别技术的研究　出处：《华南理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：汽车牌照识别技术不仅可以应用在交通管理领域，而且在收费站、停车场等车辆管理领域以及其它公共场所等民用领域也都有广泛的应用；目前随着图象识别技术的快速发展，汽车牌照识别技术的应用也已经渗透到了当今社会的各个领域，并且为社会管理和人民生活带来了极大的便利。因此，进一步开展汽车牌照识别技术研究能够极大地促进我国车辆管理公共事业的发展，对提高我国交通领域的现代化管理水平具有重要的学术研究意义和应用价值。本文在理解并掌握汽车牌照识别相关关键技术的基础上，分别从图像预处理、汽车牌照图像定位、图像分割和图象识别等方面展开了较深入的研究。论文作者的主要研究工作包括如下内容：（一）在图像预处理技术研究中，分别对图像的灰度化、降噪、图像增强以及二值化技术进行了较深入的探讨，仿真效果表明，，通过图像预处理后，汽车牌照图像的质量得到了明显的改善；（二）在车牌定位研究中，研究了基于数学形态学和纹理特征的粗定位和精确定位方法，比较了两种定位方法的仿真效果，证明了精确定位方法的有效性和准确性；（三）在车牌字符分割技术研究中，分别研究了基于联通区域和基于投影法的字符分割方法，阐述了两种方法各自的适用性，并对其分割效果进行了对比，结果表明，在本文的研究中，投影法的定位效果较好；（四）在汽车牌照字符识别技术研究中，综合了图像预处理、定位和分割各种算法的基础上，提出了基于模板匹配和基于BP神经网络的字符识别方法，并通过对33幅随机汽车牌照图像采用MATLAB仿真进行了两种识别效果的对比，结果表明，基于BP神经网络的汽车牌照字符识别方法的识别效果较为准确和可靠。
[Abstract]:The vehicle license plate recognition technology can be used not only in the field of traffic management, but also in the toll station, parking lot and other civilian areas of vehicle management areas and other public places also have a wide range of applications; at present, with the rapid development of image recognition technology, the application of license plate recognition technology has penetrated into all areas of society, and brings great convenience to people's life and social management. Therefore, the further study of vehicle license plate recognition can greatly promote the development of China's vehicle management of public utilities, it has important academic significance and application value to improve the traffic field of our country's modern management level.
On the basis of understanding and grasping the key technologies of license plate recognition, this paper launches a deep research from the aspects of image preprocessing, license plate location, image segmentation and image recognition.
(1) in the research of image preprocessing technology, we discussed the grayscale, noise reduction, image enhancement and two value technology of image respectively. The simulation results show that the quality of license plate images has been significantly improved through image pre-processing.
(two) in the research of license plate location, we studied coarse positioning and precise location method based on mathematical morphology and texture features, compared the simulation results of two positioning methods, and proved the accuracy and validity of the precise location method.
(three) research on license plate segmentation technique, were studied and Unicom region segmentation method based on projection characters, expounds the applicability of the two methods, and the segmentation results were compared. The results show that, in this study, better positioning effect projection method;
(four) in the study of vehicle license plate character recognition technology, integrated image preprocessing, positioning and segmentation algorithms based on template matching is proposed. And the method of character recognition based on BP neural network based on 33 images, and through random car license plate using MATLAB simulation were compared, two recognition results the results show that the vehicle license plate character recognition method based on BP neural network recognition effect is more accurate and reliable.

【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495;TP391.41

【参考文献】