基于中心对称LBP算子的虹膜识别改进算法研究

发布时间：2018-11-03 17:48

【摘要】：近年来,生物识别技术越来越为人们所青睐。生物识别技术是以人体自身独有的特征进行识别的技术,主要包括虹膜、指纹、掌纹、DNA以及行走的步态等。相比较于其他的生物特征,虹膜因为其自身的独特性、不变性、防伪性以及不可侵犯性等诸多优点逐渐得到学术界与企业界的重视。因此虹膜支付完全可以成为未来支付的新方向。本文以生物识别技术中的虹膜识别为主线,详细阐述了虹膜识别算法中的图像处理的主要过程,并对虹膜定位与特征提取的算法做了改进。全文所做的主要工作有以下几点:(1)概述了生物识别的含义以及其所包含的特征内容和优缺点;介绍了虹膜的结构组成以及判断虹膜识别性能的几个指标;详细陈述了本文试验中所使用过的数据库——CASIA-Iris V4。CASIA-Iris V4数据库包含有六个不同的子数据库,每个数据库的采集设备,采集对象以及采集环境都有所不同。简明扼要的对一个完整的虹膜支付系统所需要的对象以及条件做了说明。(2)在虹膜定位中,针对虹膜外边缘边界宽泛、灰度不明显等缺点提出了一种基于边界梯度增强的边缘提取算法。首先利用瞳孔灰度比较低的特点,利用二值化得图像提取虹膜内边缘,然后使用边界梯度增强算法与Canny算子相结合的方法提取虹膜外边界,使用最小二乘圆拟合的方法对提取到的内边界和外边界的进行拟合。(3)在虹膜特征提取和匹配中环节,首先对传统的LBP算子进行了说明,并且介绍了由传统LBP所衍生出来的LBP旋转不变模式和LBP等价模式两种模式。在第(2)点的基础上,本文使用改进的中心对称LBP算子进行特征提取,对中心对称LBP算子的编码方式进行探讨,改进后的中心对称LBP算子不仅能够极大的减少内存的消耗,而且显著提升特征提取的速率。在最后的匹配中,使用基于汉明距离的算法将待测图像与存储在数据库中的图像进行比对。
[Abstract]:In recent years, biometric technology has become more and more popular. Biometric recognition is based on the unique characteristics of the human body, including iris, fingerprints, palmprint, DNA and walking gait. Compared with other biological characteristics iris has been paid more and more attention by academia and business circles because of its uniqueness invariance anti-counterfeiting and inviolability. So iris payment can become the new direction of future payment. In this paper, the iris recognition in biometrics is taken as the main line, the main process of image processing in iris recognition algorithm is described in detail, and the algorithm of iris location and feature extraction is improved. The main work of this paper is as follows: (1) the meaning of biometrics, the characteristic contents, advantages and disadvantages of biometrics, the structure of iris and several indexes to judge iris recognition performance are introduced. It is described in detail that the database CASIA-Iris V4.CASIA-Iris V4 used in this paper contains six different subdatabases. The collection equipment, object and environment of each database are different. A concise description of the objects and conditions required for a complete iris payment system is given. (2) in iris location, the outer edge of the iris is broadly defined. An edge detection algorithm based on boundary gradient enhancement is proposed. In this paper, the inner iris edge is extracted by binary image, and then the outer iris edge is extracted by combining the edge gradient enhancement algorithm with the Canny operator. The method of least square circle fitting is used to fit the extracted inner and outer boundaries. (3) in the process of iris feature extraction and matching, the traditional LBP operator is first explained. The LBP rotation invariant mode and the LBP equivalent mode derived from traditional LBP are also introduced. On the basis of point (2), the improved centrosymmetric LBP operator is used to extract the features, and the coding method of the center-symmetric LBP operator is discussed. The improved centrosymmetric LBP operator can not only greatly reduce the memory consumption. Moreover, the rate of feature extraction was significantly increased. In the final match, the algorithm based on hamming distance is used to compare the image to be tested with the image stored in the database.
【学位授予单位】：江西理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】