【作者】 薛倩；

【导师】 巨永锋；

【作者基本信息】 长安大学 ， 交通信息工程及控制， 2009， 硕士

【摘要】 随着城市化步伐加快,机动车日益普及,目前许多国家都存在交通事故频发、交通拥堵情况严重等问题。如何高效的进行交通管理,已经成为世界各国关注的焦点。在这种大的背景下,伴随着计算机技术、通信技术、信息技术的飞速发展,智能交通系统(ITS,Intelligence Traffic System)也随之诞生,并且已经成为当前交通管理发展的主要方向。车辆牌照识别(LPR)系统作为智能交通系统的核心,起着非常关键的作用。汽车牌照的自动识别是计算机视觉与模式识别技术在智能交通领域应用的重要研究课题之一,是实现智能化交通的重要部分。本文针对目前车牌识别技术存在的一些问题,研究了车牌识别的各项关键技术,在分析了近年来一些典型的车牌识别算法的基础上,最终确定一系列有效的算法对车牌进行识别。车辆牌照识别系统分为车牌定位、车牌字符分割及字符识别三大部分,它的研究主要涉及到了模式识别、人工智能、计算机视觉、数字图像处理以及人工神经网络等众多的学科领域。在车牌定位方面,车牌定位是指将车牌区域从车辆图像中定位并分割出来,是车牌识别的基础。为了找到有效的车牌定位方法,本文对车牌特征信息进行了归纳总结,介绍了车牌定位前的图像预处理技术,并根据车牌区域在水平、垂直两个方向的纹理信息较其他区域都更加丰富的特征,介绍了车牌区域定位方法。在车牌字符分割方面,分析了车牌图像二值化、倾斜矫正以及字符归一化等各种算法,研究了基于字符块提取的字符分割方法。在字符识别方面,概括比较了常用的字符识别方法,对字符预处理、特征向量的提取进行了详细分析,研究了基于BP神经网络算法的字符识别方法。通过实验,识别率超过90%,证明本文采用的技术是比较成功和可行的。更多还原

【Abstract】 With the acceleration of urbanization and increasing popularity of motor vehicles, now in many countries there are frequent traffic accidents and jams. How to manage the traffic efficiently has become the focus of attention around the world. In this environment, along with the rapid development of computer technology, communications technology and information technology, the intelligent transportation system (ITS) has become the main development direction of the current traffic management. As the core of ITS, the License Plate Recognition (LPR) System plays a very important role. Focus on some problems of LPR, this paper studies the LPR key technologies. Basing on the analysis of some typical recognition algorithms, this paper finally settles down a series of effective license plate recognition algorithms. The LPR system is composed of license plate location, character segmentation and character recognition. The LPR system involves numerous discipline domains, such as pattern recognition, artificial intelligence, computer vision, digital image processing and artificial neural network etc. The purpose of license plate location is to separate the vehicle license plate from the image. In part of location, this paper first summarizes the characteristic information of license plate, then introduces a series of image pre-processing technology. The method of location is on the basis of more abundant characteristic information on the level and vertical direction of license plate than other area. In character segmentation, this article analyzes a lot of algorithms about license plate image binarization, tilt correction, character uniform. According to the character block, it studies the character segmentation method. In character recognition, this paper summaries and compares common character recognition algorithms. In addition to above, it also analyzes detailedly the method of character pre-processing, feature vector extraction and uses character recognition method on the basis of BP neural network. Through experiments, recognition rate is over 90%. It proves that technology used in this article is successful and feasible.更多还原