跨设备条件下手背静脉识别方法的研究  

DORSAL-HAND VEIN RECOGNITION UNDER CROSS-DEVICE

作　　者：王一丁[1] 苗霞 Wang Yiding;Miao Xia(College of Information Science and Technology,North China University of Technology,Beijing 100144,China)

机构地区：[1]北方工业大学信息学院,北京100144

出　　处：《计算机应用与软件》2021年第5期175-182,共8页Computer Applications and Software

基　　金：国家自然科学基金面上项目(61673021)。

摘　　要：在不同采集设备、测试环境和手背姿态等弱约束条件下,采集到的手背静脉图像存在光照对比度差异、位置偏移以及角度旋转等问题,导致图像识别精度较差。针对该问题,提出一种基于二维小波分解提取关键点和基于生物视觉生成特征描述子的手背静脉识别方法。依据手背静脉的纹理特征,使用小波分解提取其高频分量,最终选定垂直+对角分量确定更鲁棒的关键点;基于生物视觉特性,调整描述子的结构参数使其更适用于手背静脉。将该方法用于两个不同设备采集的手背静脉数据库,其识别率达到93.4%。Under the weak constraints of different acquisition equipment,test environments,and postures of hand,there are illumination difference,position offset and angle rotation in the collected dorsal-hand vein images,resulting in poor image recognition accuracy.Aiming at this problem,a dorsal-hand vein recognition method based on two-dimensional wavelet decomposition to extract key points and bio-visual feature descriptor is proposed.According to the texture characteristics of dorsal-hand vein,more robust key points were determined by using wavelet decomposition to extract its high-frequency components,and selected the vertical&diagonal components.Based on the bio-visual characteristics,the structural parameters of descriptor were adjusted to make it more suitable for the keywords dorsal-hand vein.The method was applied to the dorsal-hand vein database collected by two different devices,and its recognition rate reached 93.4%.

关 键 词：手背静脉识别 弱约束条件 二维小波分解 多方向关键点融合 生物视觉特性 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]