基于SVM的多生物特征融合识别算法  

Recognition Algorithm for Multi-Biometric Fusion Based on SVM

作　　者：刘铁根[1,2] 李秀艳[1,2] 王云新[3] 邓仕超[1,2] 

机构地区：[1]天津大学精密仪器与光电子工程学院,天津300072 [2]天津大学光电信息技术科学教育部重点实验室,天津300072 [3]北京工业大学应用数理学院,北京100024

出　　处：《纳米技术与精密工程》2011年第1期44-47,共4页Nanotechnology and Precision Engineering

基　　金：国家重点基础研究发展计划(973计划)资助项目(2010CB327802);国家自然科学基金资助项目(60627002;30770597);天津市应用基础重点项目(06YFJZJC00400);教育部博士点新教师基金资助项目(200800561020);教育部博士点基金资助项目(20090032110051);天津大学青年教师培养基金资助课题(TJU-YFF-08B47)

摘　　要：针对单生物特征识别的局限性,提出融合手背静脉和虹膜两种生物特征实现身份识别.基于尺度不变特征变换(SIFT)提取手背静脉的局部SIFT特征并对特征点进行匹配,利用特征匹配率作为手背静脉图像的相似度测度.通过Haar小波变换实现虹膜特征编码,利用加权汉明距对虹膜进行相似度测试.最后基于支持向量机(SVM)实现两种生物特征在匹配层的融合识别.利用CASIA虹膜数据库和TJU手背静脉数据库对算法性能进行测试,其等错率为0.02%,实验结果表明,该融合算法具有很高的识别性能,为生物特征识别研究提供了新思路.Aiming at the inherent limitations of unimodal biometric systems,a multi-biometric system based on hand vein and iris was proposed.The local scale invariant feature transform（SIFT） features of hand vein were extracted and matched,and the matching ratio of features between the registered image and test image was calculated as the similarity measurement.The feature code of iris was extracted from the detail images obtained using Haar wavelet transform and the weighted Hamming distances was employed to calculate the similarities between the patterns in the database and the pattern to be identified.Finally,the scores from both matching modules were sent to support vector machines（SVM）,which realized the fusion at the matching score level and the fusion recognition result was obtained.The performance of the proposed method was tested using the TJU hand vein image database and the CASIA iris database.The equal error rate（EER） of the fusion system is 0.02%.Experimental results demonstrate that the proposed fusion method has excellent recognition performance,which provides new scientific approach to further research on biometrics.

关 键 词：多生物特征 身份识别 信息融合 支持向量机 手背静脉 虹膜 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]