【作者】 彭涛；

【导师】 王绪本；

【作者基本信息】 成都理工大学 ， 地球探测与信息技术， 2007， 硕士

【摘要】 论文主要研究在实际复杂环境中汽车图像检测与汽车牌照识别中的车牌图像处理与车牌定位算法。汽车牌照识别系统(LPR)作为一个专用的计算机视觉系统，能自动检测汽车图片并从中提取出车牌号码信息。作为智能交通的一部分正越来越受到人们的重视。然而，系统在实际的复杂环境中，由于光照、天气、车速、图像背景各种因素的影响，大大限制了汽车图像检测和车牌识别的准确性和识别速度，从而大大制约了该技术的实用化推广。目前国内大部分车牌识别研究主要停留在简单试验环境下对车牌识别研究阶段，离实际环境下(主要是卡口、高速)大规模应用还有一段距离。论文从实际出发，借助大量试验，具体所做的工作包括：1．从车牌图像检测与信息处理源头入手，对比简单试验环境下车牌抓拍和实际环境下保障检测与抓拍到高质量的车辆图片。通过大量的试验与现场实际环境采集，提出了针对车牌信息识别图像采集标准，包括车牌的大小、位置、清晰度等，并提出一整套方案，通过对图像采集检测设备的选择与配置调试，保障在现场复杂环境中能够检测到符合车牌识别标准的车辆图片。2．针对卡口、高速实际现场环境复杂，车辆背景多变条件下，为了保障识别效果，根据车牌识别的特点，对车辆图片进行预处理进行研究，包括车辆图像的灰度化、图像增强、车辆图像的二值化、车辆图像的边缘检测内容。3．对车牌的定位技术进行了初步研究，提出了根据车牌图像灰度分布与变化情况下，通过模拟线横向纵向扫描而快速得出车牌区域的算法与实际效果。通过实际应用与验证，通过本文所提出的车牌检测方式与实际环境部署，得到清晰有效车辆图片，可以大大提高复杂环境下实际应用中车牌识别的准确率。更多还原

【Abstract】 In this paper, we mainly discuss the vehicle picture exploring under complex scenes and vehicle license plate locating in the vehicle license plate recognition(LPR).As a comprehensive computer vision system, the vehicle license plate recognition(LPR) can detect automobile picture and recognize the Character in the plate automatic. More and more people are attaching importance to LPR technology. But in the real and complicated situation, because of being influenced by the factors, such as light, weather, speed of vehicle, picture background etc, LPR is limited in the accuracy and recognizing speed, it restricts LPR technology to put into practice in a large scale.In our country, most research of LPR is stay in the process of experimentation, and have some work must be done to be put into practice. This thesis go through it from lots of experimentation and base on reality. The work I did include:1、I do the research from the source of LPR which is to detect and catch automobile picture in high quality. Through lots of examination and data collection in real situation, we brought forward the standard of LPR picture, it included vehicle license plate size, location, definition etc. Also we made a set of proposal that ensuring we can catch automobile picture according with the LPR standard by the choice of picture exploring device and device deploy in the real situation.2、We do research in preprocessing vehicle pictures which include images grey, media filtering, edge detection that aim to the complicate situation and variational background.3、This thesis we discuss vehicle license plate locating technology, propose a method that use simulant line scans the image to locate the area of vehicle license。A great deal of experiments are done to contrast the methods presented in the thesis with former ones from documents, and the results show that we can get clearly and effective image, and give prominence to the correction ratio.更多还原