【作者】 王海姣；

【导师】 李文举； 王新年；

【作者基本信息】 辽宁师范大学 ， 计算机应用技术， 2008， 硕士

【摘要】 随着现代高科技的发展及交通事业的需求,智能交通系统已成为人们关注的热点问题。车牌识别系统作为智能交通系统的重要组成部分,在桥梁路口自动收费、停车场无人管理、违章车辆自动记录等领域都有着广泛的应用,它的实现具有巨大的经济价值和现实意义。车牌识别系统作为一个综合的实时的计算机视觉系统,其核心技术主要包括车牌定位、字符分割和字符识别三个部分。现有的车牌识别技术多数基于灰度图像特征,而颜色作为车牌特征中的重要信息,却没有得到充分的利用。本文主要针对车牌识别技术中颜色信息的应用进行研究。在分析了国内外研究现状的基础上,本文主要完成了以下四方面工作:研究了车牌识别技术中可利用的颜色信息。提出了基于DOG模型和颜色聚类的车牌定位算法。首先根据人眼的视觉野结构,利用DOG模型进行车牌灰度图像的边缘检测;再对该图像进行中值滤波和形态学运算,通过车牌的几何和结构特征对车牌进行粗定位;当利用字符和笔画特征定位失败后,利用车牌颜色特征,进行精确定位。提出了基于颜色聚类和神经网络的车牌类型识别算法。利用基于颜色分量垂直投影的K-means粗分类进行有效区域提取,以解决光照不均和车牌褪色等问题;利用颜色聚类对有效区域进行颜色特征提取;再通过两级BP神经网络进行车牌类型识别。提出了基于YCbCr空间的特殊字符粗分类算法。根据特殊字符在Cb及Cr空间中的特征对车牌首字符进行粗分类。实验表明,本文提出的基于DOG模型和颜色聚类的车牌定位算法对光照和噪声有很强的鲁棒性;基于颜色聚类和神经网络的车牌类型识别算法有较强的实用性;基于YCbCr空间特性的特殊字符分类有助于提高汉字的识别率。更多还原

【Abstract】 With the rapid development of our country’s road transportation, the intelligent transportation system has become a hot research topic. As an important part of the intelligent transportation system, the license plate recognition system plays an important role in many fields such as intelligent transportation control system, parking lot monitor system and automatic charging system, and etc.As an integrated real-time computer vision system, the license plate recognition system mainly consists of three parts: license plate location, character segmentation and character recognition. Color information, one of the important features of the license plates, is less considered in the existing methods.Color prior knowledge-based license plate recognition method is studied in this paper. The main works are listed as follows:This paper studies how to makes full use of the color information in the license plate recognition.The license plate location algorithm based on the DOG model and color clustering is proposed. Firstly, according to the GC receptive field model, the DOG model is structured, which is used to detect the edge, secondly license plate geometrical and structural characteristics are used in the coarse location, and finally after using the gray-scale information, the color feature is fully used in the precise location.A new method based on the color clustering and the neural network to identify the license plate type is proposed. Firstly, the confidence areas are extracted by the results of the K-means algorithm, secondly color feature are obtained, and finally the license plate type is identified by the BP neural network.The special character classification algorithm is proposed, which is based on YCbCr color space to increase the recognition accuracy of Chinese character.The experiment results show that the proposed methods of license plate location algorithm and license plate type identification algorithm are robust to noises and illumination variation; the special Chinese character classification algorithm based on YCbCr color space is beneficial to improve the character recognition accuracy.更多还原