指纹图像预处理及其特征提取算法的研究与实现(可复制论文).pdf

摘要
摘要
随着计算机与信息技术的不断发展,传统的基于信物或口令方式的安全系统显得
越来越脆弱,已经不能够适应现代安全系统的需求。为了应对这些挑战,人们把目光
转向了生物特征识别技术,希望能够借助人体的生理特征或行为动作来提高身份识别
的精度。相对于其它生物特征识别技术,指纹识别具有许多独到的优点,如稳定性和
唯一性等,更重要的是它具有很高的实用性和可行性,己经被认为是一种理想的身份
认证技术,有着十分广泛的应用前景,是将来生物特征识别技术的主流。
本文主要对指纹图像预处理及其特征提取算法进行了讨论和研究。其主要研究内
容如下:
(1) 在研究和分析经典的指纹图像分割方法的基础上,实现了一种基于最大类间
方差和图像分块处理相结合的分割方法。该方法不仅具有无需根据经验来选择阈值的
优点,而且分割效果好,运算速度块。
(2) 对指纹图像的三种方向图进行了分析和比较,并将指纹的连续方向图引入方
向滤波器中,对指纹图像实现连续方向滤波。相对于传统的块方向滤波,该方法能够
获得更好的滤波效果。
(3) 构造了一种基于边缘强度的指纹图像二值化方法,并采用方向加权中值滤波
算法对二值化后的指纹图像进行滤波去噪,从而使得指纹图像的脊线结构更加清晰、
流畅。
(4) 运用数学形态学算法对指纹图像进行细化处理,并在此基础上提取指纹的细
节特征,最后针对不同的噪声采用不同的方法滤除其产生的伪特征点。


关键词:指纹图像;图像分割;连续方向滤波;二值化;细化;特征提取

I
Abstract
ABSTRACT
With the continuous development of puter and information technology,
traditional security system, based mainly on a token or password, is ing more and
more fragile and can no longer meet the demands of modern security system. To cope with
this challenge, we turn to biometrics and hope to enhance identity verification by using
human body’s physical character and behavior. The fingerprint identification is superior to
other biometrics with its special virtues, such as uniqueness and invariability. The most
important is its practicability and feasibility. It has been considered an ideal technology of
identity verification with an extensive outlook for application in the future.
In this dissertation, the algorithms of fingerprint image preprocessing and minutiae
extraction are discussed and studied. The main content of this dissertation is as follows:
(1) After investigating and analyzing the classical approaches to fingerprint image
segmentation, a new approach to segmenting the fingerprint images is realized, which is
based on the maximum between-cluster variance (MBV) and the technology of dividing
images into blocks. The experimental results indicate that this new method not only has the
advantage in choosing the threshold without the experience, but