基于二值化图像的指纹细节点精确提取方法  

Method of accurate minutiae detection from binary fingerprint image

作　　者：陈开志[1] 胡爱群[1] 

机构地区：[1]东南大学信息科学与工程学院,南京210096

出　　处：《东南大学学报（自然科学版）》2010年第3期471-475,共5页Journal of Southeast University：Natural Science Edition

基　　金：国家高技术研究发展计划(863计划)资助项目(2009AA01Z427)

摘　　要：针对指纹图像质量较差时,基于灰度图的指纹细节点直接提取方法正确率较低,提出一种基于二值化图像的指纹细节点精确提取方法.该方法首先采用方向滤波增强指纹图像,消除各种噪声,然后在二值化图像上进行脊线跟踪,直接提取细节点.脊线跟踪时,采用平滑和加权联合处理来计算脊线下一跟踪点,并提出一种利用循环扩散方法检测脊线连通性的方法,保证脊线跟踪的正确性.实验结果表明,无论图像质量好坏,该方法提取出的细节点正确率都超过80%,优于灰度图直接提取方法及其改进算法.To deal with low accuracy rates in the method of direct gray-level minutiae detection,a method of accurate minutiae detection from binary fingerprint images is proposed.First,image enhancement is employed to eliminate noise;then,ridge line following is adopted to detect minutiae from binary fingerprint images.The next point of the ridge is calculated jointly by smoothness and weighted processing.Moreover,a loop diffusion detection method is designed to guarantee the detection accuracy by checking the connectivity of the ridge line.The results of the experiments indicate that the accuracy rate of minutiae detected by this method exceeds 80% even if the quality of fingerprint images is poor.That is superior to the method of direct gray-level minutiae detection and its improved methods.

关 键 词：图像处理 指纹识别 细节点提取 脊线跟踪 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]