基于指部关联特征的多模态图像采集系统  

Multimodal Image Acquisition System Based on Finger Correlation Features

作　　者：梁爱华[1,2] 袁家政[1,2] 和青芳[1,2] 何娟[1,2] 

机构地区：[1]北京联合大学北京信息服务工程重点实验室,北京100101 [2]北京联合大学计算机技术研究所,北京100101

出　　处：《传感技术学报》2014年第2期204-207,共4页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金项目(61271369;61372148);北京高等学校青年英才计划项目(YETP1772);北京市属高等学校创新团队建设与教师职业发展计划项目(IDHT20140508)

摘　　要：针对基于单个生物特征的身份认证安全性和稳定性不足的问题,设计了基于指部关联特征的多模态图像采集系统,采用单个双波段摄像头分时采集同一根手指的指纹、指节纹和指静脉图像。指纹和指节纹采用非接触反射采集方式,指静脉采用单侧近红外光源与反射镜面相结合的透射采集方式,并根据静脉图像质量评价动态调控光源,根据特征点信息量动态调整各个特征的权重。实验结果表明,该多模态采集系统在认证通过率、误识率和拒登率等指标都优于指纹或指静脉的单模态采集系统,认证通过率达到99.1%,误识率为0.000 1%,不存在拒登现象。To avoid the low security and instability of single biometric authentication,a multimodal image acquisition system based on finger features is designed to realize the time-sharing collecting the fingerprint,finger knuckle and finger-vein images through a dual-band camera. Fingerprint and finger knuckle images are acquired by non-contact reflection way. The single light source and reflection mirror is adopted while collecting finger-vein image. The near-infrared LED light source position and angle can be adjusted according to the finger-vein image quality assessment. The weights are adjusted based on the information contents of three features. Actual test results show that multimodal image acquisition system has distinctive advantages compared to single-mode image acquisition system,which was based on fingerprint or finger-vein. In proposed system, authentication passing rate can reach 99. 1% and false accept rate is 0. 000 1% without rejection registration.

关 键 词：图像采集 多模态 特征融合 指纹 指节纹 指静脉 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]