非接触式掌静脉识别研究

发布时间：2018-04-01 03:05

本文选题：掌静脉识别　切入点：OTSU　出处：《北京交通大学》2017年硕士论文

【摘要】：掌静脉识别技术是通过分析手掌里的静脉血管信息进行身份识别的一项生物特征识别技术,具有识别准确、活体认证、易于接受等优点,近年来得到了广泛的关注,并成功地应用于安检、金融、社保资金发放等领域。目前对于掌静脉识别技术的研究大多集中于对特征提取算法的研究,对于掌静脉图像获取之后的预处理算法研究的并不多,然而对于非接触式采集掌静脉图像,每次采集时手掌的位置、方向以及手指的张开角度都会有所不同,因此预处理的结果将直接影响到最后识别的准确性。为解决上述这些问题,本文首先对掌静脉图像的预处理进行了比较详细的研究和讨论,然后提出了一种新的掌静脉特征提取算法和匹配算法,取得了比较满意的识别效果。论文主要研究工作如下:(1)针对传统的掌静脉识别算法没有考虑获取感兴趣区域(Region Of Interest,ROI)的各种复杂情况,而非接触式采集到的掌静脉图像引入的噪声等各种不可控因素较多,本文对预处理算法进行了系统的定量研究——对基于固定长度和相对长度的ROI获取进行了对比实验。(2)提出了一种新的基于主曲率的掌静脉特征提取算法。首先对ROI进行梯度标准化,去除干扰噪声;然后求取图像的最大主曲率,将掌静脉结构同周围的人体组织区分出来;最后再对图像进行分割腐蚀提取出了清晰的掌静脉特征。(3)提出了基于模版匹配的掌静脉特征匹配算法。在掌静脉图像的特征匹配阶段,针对主曲率提取出的掌静脉特征,使用模版匹配法进行识别,并且和LBP匹配法进行对比。实验证明ROI的边长取1.5倍左右的食指、中指间的谷点和无名指、拇指间的谷点之间的距离时,可以达到最好的识别效果,并且主曲率法结合模板匹配法可以达到1.965%的等误率(Equal Error Rate,EER),识别效果准确。
[Abstract]:Metacarpal vein recognition is a biometric identification technology based on the analysis of venous vascular information in the palm. It has the advantages of accurate identification, living authentication and easy to accept, and has been paid more and more attention in recent years. And it has been successfully applied in the fields of security, finance, social security fund distribution and so on. At present, most of the researches on palmar vein recognition are focused on the feature extraction algorithm, and there are few researches on the pre-processing algorithm after the palmar vein image acquisition. However, for non-contact acquisition of metacarpal vein images, the position of the palm, the direction of the palm and the angle of the opening of the fingers are different with each acquisition. Therefore, the result of preprocessing will directly affect the accuracy of final recognition. In order to solve these problems, this paper firstly studies and discusses the preprocessing of metacarpal vein image in detail. Then a new palmar vein feature extraction algorithm and matching algorithm are proposed. The main work of this paper is as follows: 1) for the traditional palmar vein recognition algorithm, the region of interest (region of interest) is not taken into account. However, there are many uncontrollable factors such as noise and other uncontrollable factors in the non-contact image of palmar vein. In this paper, a systematic quantitative study of preprocessing algorithm is carried out. A new algorithm for extraction of metacarpal vein features based on main curvature is proposed. Firstly, the ROI is extracted by comparing the ROI acquisition based on fixed length and relative length. For gradient standardization, The interference noise is removed, the maximum principal curvature of the image is obtained, and the palmar vein structure is distinguished from the surrounding human tissue. Finally, a palmar vein feature matching algorithm based on template matching is proposed. In the phase of metacarpal vein image feature matching, the palmar vein feature is extracted from the principal curvature. The template matching method is used to identify and compare with the LBP matching method. The experiment proves that the best recognition effect can be achieved when the side length of ROI is about 1.5 times that of index finger, the distance between middle fingers and ring fingers, and the distance between thumbs and valley points. The principal curvature method combined with template matching method can achieve equal Error error rate of 1.965%, and the recognition effect is accurate.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】