光学学报, 2018, 38 (4): 0411007, 网络出版: 2018-07-10
一种存储伪图像的手掌静脉识别研究 下载: 808次
Palm Vein Recognition with Pseudo Image Storage
机器视觉 手掌静脉识别 生物特征模板保护 近红外光图像 伪图像 machine vision palm vein identification biometric temple protection NIR image pseudo image
摘要
传统的生物特征识别方法直接将用户的生物特征模板存储于数据库中。由于生物特征具有唯一性和稳定性,一旦被窃取,用户的生物特征将终身不可再用。针对此问题提出了一种存储伪生物特征图像的手掌静脉识别算法,该算法不从外部输入密钥,不存储原始掌脉生物特征模板。在注册阶段,于近红外光下采集用户手掌静脉图像,对图像进行加密形成伪图像,将伪图像存储于数据库中;在认证阶段,将数据库中的伪图像解密后提取特征,与认证阶段采集图像提取的特征进行匹配,给出认证结果。在PolyU图库、CASIA图库和自建图库上进行测试,结果表明:在样本数量为300时,该算法在上述3种图库中的等误率分别为0.4135%、0.5576%、0.4744%,识别时间分别为325.0740,316.0800,322.6530 ms。在小范围样本内,所提算法适用于安防、考勤等场合,具有一定的实用价值。
Abstract
Traditional biometric system stores the biometric template in the database directly. While the biometric template is unique and permanent, it can’t be used anymore if stolen. To solve this problem, we propose a palm vein recognition algorithm with pseudo image storage, which doesn’t input any key information and doesn’t store original palm vein template. In the stage of register, the system collects palm vein images under near-infrared light. Then the collected image is stored in the form of pseudo image. In the stage of recognition, the pseudo image is decrypted and the feature is extracted. The feature is matched with input image and the authentication result is given. The proposed algorithm is tested on PolyU database, CASIA database, and self-built database. The experimental results show that in sample range of 300, the proposed algorithm can reach the equal error rates of 0.4135%, 0.5576%, and 0.4744% and recognition times of 325.0740 ms, 316.0800 ms, and 322.6530 ms for the above databases. The algorithm has practical value for security and checking-in occasions in certain sample range.
吴微, 林森, 苑玮琦. 一种存储伪图像的手掌静脉识别研究[J]. 光学学报, 2018, 38(4): 0411007. Wei Wu, Sen Lin, Weiqi Yuan. Palm Vein Recognition with Pseudo Image Storage[J]. Acta Optica Sinica, 2018, 38(4): 0411007.