【作者】 赵晶；

【导师】 高隽； 张旭东；

【作者基本信息】 合肥工业大学 ， 信号与信息处理， 2004， 硕士

【摘要】 统计学习理论是一种专门研究小样本情况下机器学习规律的理论，支持向量机方法是在该理论基础上发展起来的通用学习方法，它具有全局优化、适应性强、理论完备、泛化性能好等优点，统计学习理论和支持向量机是目前机器学习领域的研究热点。随着通信、信息和电子技术及计算机网络技术的发展，智能交通系统正越来越受到各国的重视，它包括车型识别、车牌识别等模块。 本文将支持向量机引入智能交通系统领域，主要进行的工作如下： (1) 整理总结了国内外学术界关于统计学习理论方面的研究成果，介绍统计学习理论的基本概念和支持向量机的基本原理； (2) 在形状识别问题中以交通标志图像作为实验对象，利用Hough变换进行特征提取，在识别阶段利用支持向量机方法进行分类，并与神经网络等传统学习方法对比； (3) 将支持向量机应用于车型识别问题中，针对收费站采集的汽车图像，首先采用小波分析和数学形态学的方法提取其外形特征，在识别阶段利用支持向量机方法进行分类，并与其他传统学习方法进行了对比； (4) 将支持向量机应用于车牌识别问题中，车牌识别包括车牌定位、车牌字符分割以及字符识别三个步骤，先采用数学形态学方法对车牌区域进行定位，然后采用Top-Hat变换等方法分割车牌字符，在识别阶段采用支持向量机算法进行字符识别，取得了较为满意的结果。更多还原

【Abstract】 Statistical Learning Theory(SLT) is a learning theory which specializes in machine learning with finite examples. As a learning method based on SLT, Support Vector Machine(SVM) has the advantages of global solutions, good adaptivity, high generalization ability and maturity in theory. SLT and SVM are the hot-spot in the field of machine learning nowadays. With the development of communication, information and electronic technology and computer network, Intelligent Transport System(ITS) is paid more and more emphasis, it contains many parts, such as vehicle type recognition and license plate recognition.In this paper, we introduce SVM to the field of ITS, the main work is described as follows:(1) We summarize the latest research achievements and development of ITS, present the conceptions of SLT and the principles of SVM;(2) Taking the traffic sign as examples and adopting Hough transform in the stage of feature extraction, we introduce SVM to the problem of shape recognition and compare the experimental results with traditional learning methods.(3) Then we use SVM to settle the vehicle type recognition problem, where we utilize the wavelet analysis and mathematical morphology method to extract the figure feature. In the stage of test, the generalization ability of SVM exceeds other methods.(4) Lastly, SVM is used here in the problem of license plate recognition, which contains license plate location, characters segmentation and character recognition. In this paper, we make use of morphology method to locate the region of license first, and then top-hat transform is used here to segment the characters, in the stage of character recognition we adopt SVM method and the result is satisfying.更多还原