汽车牌照识别系统毕设毕业论文.doc

中文摘要
随着城市交通管理现代化水平的提高,建立在车辆牌照识别基础上的交通信号自适应控制系统、智能交通监控系统、GPS车辆管理及导航系统、智能化交通管理系统应运而生。由于汽车牌照是机动车辆管理的主要标志符号之一,因此对车辆牌照识别系统的研究就尤为重要,该研究的核心是提高车牌识别准确性,这就需要识别算法能够对环境光照条件,拍摄位置和车辆行驶速度等因素的影响有较大的鲁棒性,并能够满足实时性的要求。
本文设计的主要内容包括:运用MATLAB仿真软件对后台图像进行高速,准确的处理。主要工作是利用BP神经网络算法对牌照的字符识别进行了研究。在牌照的定位之前,本文首先运用不同于当下传统的方法,而是结合当前比较热门的BP神经网络来对图像进行预处理,得到了比较突出的牌照信息。然后结合牌照自身的一些固有特征,用形态学的方法进行了车牌区域的准确定位;对定位得到的车牌,再次结合牌照内部细节特征,对车牌进行投影,从而得到了车牌较为完整的分割;最后在车牌字符识别阶段,以现今比较流行的神经网络为理论基础,在原有的神经网络算法基础上进行有效改进。仿真结果表明本算法在车牌识别中具有明显的优势。
本文研究内容的创新性体现在以下三个方面:(1)车牌定位阶段,N神经网络对图像进行预处理可以得到一个细节丰富,边缘完整的二值图像,随后基于牌照固有特征,结合一种新的形态学方法,可以得到多个车牌候选区域:(2)车牌字符分割阶段,为了满足算法实时性的要求,对畸变车牌进行预处理时,采用字符的行特征来描述本文的倾斜校正算法,之后,基于改进的投影特征图对字符图像分割;(3)字符识别阶段,为了提高算法的准确性,采用改进的BP神经网络。
关键词:数字图像处理车牌识别字符识别神经网络
An Arithmetic Research Based On work for Car License Plate Recognition
Abstract
With the rapid development of domestic traffic management, these systems that are based on the license plate recognition appeared at the right moment, such as the traffic signal automation management system, intelligent traffic monitor system, GPS, automation parking toll system, and intelligence traffic system, . Because license plate is a important part of vehicle management system, so the research of license plate system is particularly important .The core of arithmetic of license plate system is how to improve the recognition, so it requires recognition algorithm has great robustness for the impact of light conditions of the environment and taken the position and vehicle speed it also can satisfy the timely requirement.
In this paper, my work is focused on the image processing based on MATLAB emulator. Three problems were analyzed, which are license locating, segmentation and character recognition. Before license locating, artificial work (ANN) is adopted to handle the image, then the information of license plate is obtained in evidence. To locate the license plate, the inherent features of license plate being used, and then use mathematical morphology to locate th