面向智能停车场的车牌识别技术研究

发布时间：2019-01-17 08:54

【摘要】：车牌识别技术作为智能交通管理的核心技术,已广泛应用于道路交通管理、智能停车场出入口等领域,大大提高了车辆管理效率,实现车辆的科学化和智能化管理,有着广泛的应用前景。本文开发了智能停车场管理系统,重点研究了车牌自动识别技术,提出对车牌图像进行超分辨率重建以提高车牌图像质量,借鉴前人的车牌定位方法和字符分割方法进行车牌定位和字符分割,提出一种改进的模板匹配字符识别方法,并用Matlab进行仿真实验,最后进行系统测试和结果分析。主要研究内容包括:(1)采集到的车牌图像质量好坏影响着车牌定位和识别效果。本文分析和比较了各种超分辨率图像重建方法,选择了凸集投影超分辨率图像重建方法,研究其理论基础和算法实现步骤。实验结果表明,重建后的车牌图像,分辨率明显提高,图像质量更高。(2)分析和比较了现有的各种车牌定位方法,本文借鉴和参考前人的基于形态学的灰度化图像车牌定位方法;分析和比较了现有的字符分割方法的优缺点,借鉴和参考前人的基于投影法和连通元特征结合的字符分割方法。利用Matlab进行车牌定位和字符分割的仿真实验。(3)综合分析和比较了现有各种字符识别方法,对传统模板匹配字符识别方法进行算法改进,先根据字符特征进行字符粗分类,再进行模板匹配,实验表明该方法对于相似字符的识别效果好。(4)开发智能停车场管理系统,对本文介绍的车牌识别技术进行测试。仿真实验结果发现,超分辨重建可以提高图像分辨率和图像质量,最终提高系统准确率,采用本文改进的模板匹配字符识别算法,汉字识别准确率为91.71%,数字和字母识别准确率为95.29%,系统整体测试的识别准确率为90.76%。实验数据表明,本文介绍的车牌识别方法准确率较高,复杂环境下表现好,具有较好的实用性。
[Abstract]:As the core technology of intelligent traffic management, license plate recognition technology has been widely used in the fields of road traffic management, intelligent parking lot entrance and exit, and has greatly improved the efficiency of vehicle management and realized the scientific and intelligent management of vehicles. It has a wide application prospect. In this paper, the intelligent parking lot management system is developed, and the automatic license plate recognition technology is studied, and the super-resolution reconstruction of license plate image is proposed to improve the quality of license plate image. Based on the previous license plate location method and character segmentation method, an improved template matching character recognition method is proposed. The simulation experiment is carried out with Matlab. Finally, the system test and result analysis are carried out. The main research contents are as follows: (1) the quality of the collected license plate image affects the license plate location and recognition. This paper analyzes and compares various super-resolution image reconstruction methods, selects convex set projection super-resolution image reconstruction method, and studies its theoretical basis and algorithm implementation steps. The experimental results show that the resolution of the reconstructed license plate image is obviously improved and the image quality is higher. (2) the existing license plate localization methods are analyzed and compared. This paper draws lessons from and refers to the former method of license plate location based on morphology in grayscale image. The advantages and disadvantages of the existing character segmentation methods are analyzed and compared, and the character segmentation methods based on the combination of projection method and connected element feature are used for reference. The simulation experiment of license plate location and character segmentation using Matlab is carried out. (3) the existing character recognition methods are comprehensively analyzed and compared, and the traditional template matching character recognition method is improved. The experiment results show that the method is effective for the recognition of similar characters. (4) the intelligent parking lot management system is developed and the license plate recognition technology introduced in this paper is tested. Simulation results show that super-resolution reconstruction can improve image resolution and image quality, and finally improve the accuracy of the system. Using the improved template matching character recognition algorithm, the recognition accuracy of Chinese characters is 91.71. The recognition accuracy of numerals and letters is 95.29 and the recognition accuracy of the whole system is 90.76. The experimental data show that the license plate recognition method introduced in this paper has high accuracy, good performance in complex environment and good practicability.
【学位授予单位】：华侨大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U495;TP391.41

【相似文献】