用于特征提取的静脉图像质量评价方法  

Vein image quality assessment method for feature extraction

作　　者：黄忠主[1] 穆春迪[1] 谢剑斌[1] 刘通[1] 陈章永 

机构地区：[1]国防科学技术大学电子科学与工程学院,湖南长沙410073 [2]中钞实业有限公司,北京100052

出　　处：《计算机工程与科学》2013年第10期186-190,共5页Computer Engineering & Science

基　　金：国家863高技术研究发展计划资助项目(2009AA04A101);公安部创新计划资助项目(2010YYCXHNST064)

摘　　要：针对手指静脉图像评价速度与效果的要求,提出了一种基于区域列平均梯度特征与纹理粗糙度特征相结合的手指静脉图像质量快速评价方法。该方法首先定义区域列平均梯度特征,并依据此特征进行一次质量判决,满足要求则进入手指注册阶段;否则再利用纹理粗糙度特征进行二次质量判决,满足要求则进行手指注册,否则提示用户更换手指进行注册。仿真实验表明,本文方法评价效果好,速度快,能够满足手指静脉识别的要求。In order to achieve quick and reliable assessment of finger vein image,the paper proposes an assessment method of vein image quality based on the features of regional column average gradient and texture roughness.In this method,firstly,the feature of regional column average gradient is defined on the basis of which the first quality assessment is made.If the requirement is met,finger registration will be done directly.Otherwise the second assessment is conducted based on the feature of texture roughness.If the requirement is met this time,finger registration will be done.If not,the system prompts the user for another finger to register.The results of simulation experiment show that this method is effective and efficient and it is able to meet the requirement of finger vein recognition.

关 键 词：区域列平均梯度 纹理粗糙度 手指静脉识别 图像质量评价 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]