车牌号识别方法研究

发布时间：2019-05-18 12:00

【摘要】：随着我国经济的不断发展,汽车的数量急剧增加,智能交通系统也就越来越重要,因此车牌号识别作为智能交通系统的重要组成部分,也就有更高的研究价值和实用价值。本文主要对车牌号识别的相关算法进行研究,本文车牌号的识别过程主要包括:图像预处理、车牌定位、车牌分割、字符识别四个部分。通过对我国车牌特点的了解,本文对车牌号识别过程中用到算法进行了研究,分析和比较了各个模块采用的算法的优点和缺点,将不同算法的优点结合起来,使其更好的应用在车牌号识别的过程中。首先对采集的车牌图像进行预处理,依次是灰度化、边缘检测、二值化;在车牌定位模块中,采用基于数学形态学和投影的算法进行车牌定位,并且在数学形态学定位后,采用基于Hough变换的算法对车牌进行倾斜校正;在车牌分割模块中,比较和分析了基于模版的车牌分割算法和基于投影的车牌分割算法的优缺点,并将两种方法的优点结合应用在车牌字符进行分割中,实现了字符分割;在车牌字符识别模块中,研究了基于模版匹配的字符分割的算法的优缺点,改进了模板库的建立方法和字符匹配的方法,采用加权匹配的方法,提高了字符识别的准确率和速度,识别结果正确率为92%,识别时间为500ms左右;最后,在车牌识别系统中实现了各个模块中用到的算法,并且构建了一个识别界面,显示并存储车牌识别过程中每一步的结果。
[Abstract]:With the continuous development of economy in our country, the number of cars increases sharply, and the intelligent transportation system is becoming more and more important. Therefore, as an important part of intelligent transportation system, license plate number recognition has higher research value and practical value. This paper mainly studies the related algorithms of license plate number recognition. the recognition process of license plate number mainly includes four parts: image preprocessing, license plate location, license plate segmentation and character recognition. Through the understanding of the characteristics of license plate in our country, this paper studies the algorithm used in the process of license plate number recognition, analyzes and compares the advantages and disadvantages of the algorithms used in each module, and combines the advantages of different algorithms. Make it better used in the process of license plate number recognition. Firstly, the collected license plate image is preprocessed, followed by grayscale, edge detection and binarization. In the license plate location module, the algorithm based on mathematical morphology and projection is used to locate the license plate, and after the mathematical morphology location, the algorithm based on Hough transform is used to correct the tilt of the license plate. In the license plate segmentation module, the advantages and disadvantages of the template-based license plate segmentation algorithm and the projection-based license plate segmentation algorithm are compared and analyzed, and the advantages of the two methods are combined in the license plate character segmentation, and the character segmentation is realized. In the license plate character recognition module, the advantages and disadvantages of the character segmentation algorithm based on template matching are studied, and the establishment method of template base and the method of character matching are improved, and the weighted matching method is adopted. The accuracy and speed of character recognition are improved, the correct rate of recognition result is 92%, and the recognition time is about 500ms. Finally, the algorithm used in each module is implemented in the license plate recognition system, and a recognition interface is constructed to display and store the results of each step in the license plate recognition process.
【学位授予单位】：中国石油大学(华东)
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;TP391.41

【参考文献】