基于局部联合边缘和方向模式的多光谱掌纹融合识别  

Multispectral Palmprint Fusion Recognition Based on Local Joint Edge and Orientation Patterns

作　　者：李梦雯 施汉琴[1] LI Meng-wen;SHI Han-qin(School of Computer Science and Technology,Huaibei Normal University,Huaibei 235000,China)

机构地区：[1]淮北师范大学计算机科学与技术学院,安徽淮北235000

出　　处：《长春师范大学学报》2020年第6期69-80,共12页Journal of Changchun Normal University

基　　金：安徽省高等学校自然科学研究一般项目“基于鲁棒二值特征学习的多光谱掌纹融合识别方法研究”(KJ2019B02)。

摘　　要：针对掌纹图像边缘梯度特征和方向特征较突出的特点,提出一种新的局部联合边缘和方向模式,用于提取多光谱掌纹特征。首先,采用Kirsch算子计算掌纹图像在8个不同方向的边缘响应值,并提出局部最大边缘模式描述边缘特征。采用Gabor滤波器或改进的有限Radon变换提取掌纹图像的方向特征。然后对边缘特征和方向特征进行联合分析,构建二维特征矩阵。在MSpalmprint掌纹库上的实验结果表明,与现有的掌纹识别方法相比,基于局部联合边缘和方向模式的方法具有更高的识别率、更低的等错误率和更快的识别速度。最后,利用求得的等错误率对各光谱下的匹配分数进行加权融合。其中,对Green、Red和NIR三种光谱的掌纹进行融合识别后的等错误率最低为0.0001%。In view of the prominent edge and orientation feature in palmprint images,a new local joint edge and orientation patterns(LJEOP)is proposed to extract the multispectral palmprint features.Firstly,the Kirsch operator is used to calculate the edge response of the palmprint image in eight different directions,and the local maximum edge patterns(LMEP)is proposed to characterize the edge feature.The Gabor filter or the improved finite Radon transform(MFRAT)is used to extract the orientation feature.Secondly,the two-dimensional feature matrix is constructed based on a combination of edge and orientation analysis.Experimental results on MSpalmprint database show that the LJEOP method has higher recognition rate,lower equal error rate and faster recognition speed than state-of-the-art methods.Finally,the matching scores of each spectrum are weighted by using the equal error rate.The best result of fusion is obtained by fusing Green,Red and NIR bands,which has EER as low as 0.0001%.

关 键 词：多光谱掌纹 KIRSCH算子 GABOR滤波器 改进的有限Radon变换 分数层融合 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]