基于视频图像的车辆检测和车牌识别

发布时间：2018-01-12 22:08

本文关键词：基于视频图像的车辆检测和车牌识别　出处：《宁夏大学》2017年硕士论文　论文类型：学位论文

【摘要】：车牌识别系统是智能交通系统(Intelligent Transport System,简称ITS)的一个重要组成部分,车牌识别在城市交通管理、停车场管理、小区车辆出入、高速公路收费等方面有着广泛的应用,对于城市道路交通卡口系统而言,高效的对车辆进行检测、对车牌进行识别是协助破获肇事逃逸、车辆违章和车辆丢失等事件的有效手段。在车牌识别系统中,能否检测到运行车辆并对运行车辆进行准确的车牌定位对车牌字符的识别有着极其重要的影响,不能准确的定位到车牌就不能进行字符识别,因此本文会重点对车辆检测和车牌定位进行深入研究,并提出一种新的车辆检测和车牌定位方法,并在Matlab平台上编程实现,实验结果表明了该方法的有效性。本文的主要研究内容包括以下几个方面:运动车辆的检测。分析了目前常用的一些前景提取方法的优缺点,在Matlab上分别编程实现其中三种提取车辆的方法,对比实验结果并分析。根据实际需要选择背景差分法的平均值背景模型并结合车辆的梯度图像检测目标车辆,最后使用形态学方法和二值化方法提取完整的目标车辆。车牌定位。简要介绍目前针对车牌定位常用的一些方法,分析它们的优缺点,通过实验验证一些车牌定位方法存在的缺点,例如,边缘检测方法对于图像质量较差的车辆图片的车牌定位存在很大的误差。根据这些缺点提出一种结合边缘检测和颜色特征的车牌定位方法,并进行实验验证该方法的可靠性。车牌倾斜校正与字符分割。成功定位到车牌后需要为车牌识别做准备工作,定位到的车牌往往存在倾斜和定位的精确度不够的问题,根据这些问题选择一些算法校正倾斜车牌,校正后的车牌根据垂直投影法进行精确定位并分割车牌字符。字符识别。分析目前现有的车牌字符识别算法,结合实际情况选择合适的算法识别车牌字符。
[Abstract]:License plate recognition system is the intelligent transportation system (Intelligent Transport System, referred to as ITS) is an important part of license plate recognition in city traffic management, parking lot management, vehicle access, is widely used for highway tolls, city road traffic monitoring system, efficient detection of the vehicle, the license plate recognition is to assist cracked escapes, effective means of vehicle peccancy and vehicle loss events. In the license plate recognition system can detect vehicle operation and the operation of the vehicle license plate accurately is very important for the license plate character recognition, can not accurately to locate the license plate character recognition is not so. This paper will focus on in-depth study of vehicle detection and license plate location, and proposed a new vehicle detection and license plate location method, and in Matlab The realization of programming platform, the experimental results show that the method is effective. The main contents of this paper include the following aspects: vehicle detection. Analyzed the advantages and disadvantages of some commonly used foreground extraction method, respectively in Matlab programming method of the three extraction vehicle, and comparative analysis of the experimental results. According to background difference method the average background model based on the detection of target gradient image of the vehicle actual need, the target vehicle finally using morphological method and binarization method for extracting intact. The license plate location. This paper briefly introduces some common methods for locating and analyzing their advantages and disadvantages, through experimental verification of some license plate the positioning methods, for example, the edge detection method for vehicle license plate location picture of poor image quality in the presence of large errors. According to these disadvantages A method combining edge detection and color feature of license plate location method, and experiments are carried out to verify the reliability of the method. The license plate tilt correction and character segmentation. The successful positioning of the license plate to do the preparatory work for license plate recognition, license plate location tend to tilt and positioning accuracy is not enough, some selection algorithm according to the problem vehicle license plate tilt correction, after correction for precise positioning according to the vertical projection and segmentation of license plate characters. Character recognition. Analysis of the existing license plate recognition algorithm, combined with the selected algorithm of license plate character recognition of the actual situation.

【学位授予单位】：宁夏大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495;TP391.41

【参考文献】