【作者】 庄大燕；

【导师】 王科俊；

【作者基本信息】 哈尔滨工程大学 ， 模式识别与智能系统， 2006， 硕士

【摘要】 信息时代的到来给人类社会带来了极大的变革，信息安全显得尤为重要，身份认证在信息时代也有了新的内涵。利用生物特征的身份认证由于利用了人体固有的生理和行为特征而展现出极大的优越性。目前，一些生物特征识别技术已经很成熟而且应用广泛，比如指纹识别，而一些生物特征识别技术的研究还处于起步阶段，比如静脉识别。 本文就人体手背静脉识别作了具体的研究工作。首先分析了静脉识别的理论依据和它相对于其它生物特征识别的一些优点。在具体的研究工作中总结出手背静脉识别在获取符合要求的手背静脉图像、手背静脉图像分割和手背静脉图像特征提取以及匹配算法这三个方面难度比较大，本文从人体手背静脉识别的三个难点出发，研究出一套比较理想的人体手背静脉识别系统。 在静脉图像处理模块中，对获得的静脉图像进行预处理，用多种动态阈值法二值化静脉图像，并比较分析动态分割效果，提出了一种新的适合静脉图像的图像分割方法——基于边界的图像分割。在特征提取与匹配中，本文试验并新提出了很多方法，其中包括基于统计特性的K-L变换的方法。提出了一种基于特征点的新的特征匹配方案，即先用星型结构进行局部匹配去除伪特征点，然后按照距离匹配进行全局匹配，实验证明结果很好。同时提出了一种基于特征矩的模板匹配方法，也取得了很好的识别结果。以上各种方法各有自己的优缺点，因此本文提出了一种基于决策层的结果融合的静脉识别新方法，避免了由单一特征识别带来的误差。最后提出了自己的关于静脉识别研究的一些想法。更多还原

【Abstract】 There are great changes happened in the human society following the advent of the information age. Information safety is looked more important. And the identity authentication has new connotation in the information age. The identity authentication technology based on biological characteristic has showed a great of advantages because of the use of inherent physiology and behavior characteristics of human. At present, some biological feature recognition technologies are very mature and have been using extensively, such as fingerprint recognition, and the research of some biological feature recognition technologies are also at the starting stage, such as the hand vein recognition.In this paper, the hand vein recognition is studied deeply. First, the theoretical foundation of hand vein recognition and some advantages of it compared to other biological feature recognitions. In the study work we found that there are three difficulties in hand vein recognition. They are obtaining the fulfilling requirements hand vein images, the threshold segmentation of the hand vein images and the feature extraction and matching method of hand vein images. In this paper, an ideal suit of hand vein recognition system is developed proceeding from these three difficulties of the hand vein recognition.In the hand vein image processing, preprocess the hand vein image first. Several dynamic threshold segmentation methods are used to segment hand vein image and the dynamic segmentation results is compared, a kind of new dynamic threshold segmentation is proposed which is suitable for the vein images -the Method of Image segmentation based on boundary. In the part of feature extraction and matching, many methods are tried, including the K-L transform method. A new matching method based on feature points is developed. The false feature points are got rid of first using the star model structure, then global matching is done based on the distances method. It came更多还原