大庆石油学院应用技术学院毕业论文摘要生物特征识别技术因其在身份识别时的准确、方便、不易仿造等特点，正在被越来越多的学者和研究机构所重视。所谓生物特征识别技术是指通过计算机利用人体所固有的生理特征或行为特征来进行个人的身份识别。目前，已有很多生物个体特征识别技术被研究和使用，但在系统的准确性、方便性和实用性等方面，它们尚不能满足人们的要求。（上段以引言的方式简要说明论文立题的背景、意义和进行这方面研究的必要性。）本文首先分析比较了各种生物特征在身份识别的不同指标下所表现的能力，从方便性、准确性和实用性的角度出发，提出了采用人脸特征和语音信息相结合的组合特征方法,并通过对动态和静态两种特征的研究，完成它们的特征提取和有效的数据融合，更为全面的将人的个体特征信息提供给识别系统。同时本文针对生物特征识别技术中存在的问题，对神经网络技术及其在模式识别系统中的应用进行了深入的研究。首先，针对特征数据量大、各种数据对特征的表征能力不同的问题，提出采用APEX网络实现特征优化；其次，针对神经网络构建时，因网络结构和参数初始化无通用方法可寻而采用随机赋值造成网络收敛缓慢的问题，提出了一种新的通用的神经网络构建方法WKBNN；最后，针对开机身份识别的特点，以及神经网络对在完成多类别分类情况下的训练和识别时网络收敛缓慢的问题，提出了采用子网组合方式来设计识别器。作为这些研究成果的应用，本文在最后用Matlab程序实现了一个基于神经网络的生物特征身份识别系统试验平台。并根据对参试者进行的大量身份测试试验，总结系统的各方面能力和分析存在的问题，为进一步的研究提供了方向和宝贵的经验。（上面这部分扼要介绍作者在论文中陈述的主要内容）关键词神经网络；模式识别；数据融合；生物特征；人脸识别I目录目录摘要······································································································I第1章章标题·······················································································11.1节标题···························································································11.1.1小标题······················································································11.1.2小标题······················································································21.2节标题···························································································31.2.1小标题······················································································31.2.2小标题······················································································41.2.3小标题······················································································51.2.4小标题·········································································...
VIP
