一种二步分类计算指纹纹线距离的新方法  

A New Method Based On Two-Step Classfication For Ridge Distance Estimation Of Fingerprint Image

作　　者：谭薏[1] 马志庆[1] 刘飞[1] 徐公明[1] 

机构地区：[1]山东中医药大学理工学院,济南250355

出　　处：《微计算机信息》2010年第19期190-192,共3页Control & Automation

摘　　要：纹线距离是指纹图像纹理结构的根本属性,在自动指纹识别中有着重要的应用。现有的纹线距离计算方法往往忽略指纹局部质量差异对纹线距离估计的影响,而直接对指纹纹线距离进行同一化计算,造成了纹线距离计算失准。本文针对该问题,提出了一种两步分类计算指纹纹线距离的方法。实验结果证明该方法显著提高了算法对指纹图像质量的鲁棒性。Fingerprint ridge distance is the fundamental trait of fingerprint image texture and plays an important role in automatic fingerprint identification. The key methods of calculating ridge distance often overlook the quality of the fingerprinand and calculate the global image identically,as a result,fingerprint ridge distance is computed inaccuratly. To deal with this problem,a new two -step method based on the fingerprint quality classification is presented. First of all,the entropy theory is induced in order to evaluate the global fingerprint image quality,and then,according to the quality result,the good and poor fingerprint images are separated. For the selected good fingerprint images,the Large Window method based on spectrum analysis is employed;for the poor fingerprint image,the quality factor is employed to choose the adjoined good blocks （local representive connected region）,and then,the Statistical Window method is employed to calculate local representice fingerprint distance,which results in the final. The experimental result showes that this method can improve the robustness of the algorithm for fingerprint quality significantly.

关 键 词：指纹 熵理论 统计窗 纹线距离 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]