可撤销手掌纹脉二维PalmHash码的共轭认证

Conjugate verification of cancelable palm-print-vein using 2D-PalmHash code

由于认证性、实用性和防伪性方面的多重优势,迫切需要开发融合掌纹和掌脉的双模态认证技术。作为新型的可撤销掌纹认证编码,二维PalmHash码(2DPHC)克服了直接使用原始掌纹特征的安全隐患和隐私问题。本文将2DPHC框架推广至可撤销掌脉,通过融合掌纹和掌脉的2DPHC,实现可撤销手掌纹脉认证。为了不增加双模态认证中的计算复杂度和模板数据量,对掌纹和掌脉的模板生成方向进行优选。在每种模态中,分别生成4个方向的2DPHC,并统计匹配分数协方差。根据小协方差选择策略,分别从掌纹和掌脉中选择低相关性的2个2DPHC作为可撤销共轭模板。通过均值策略融合多个模板的匹配分数,实现可撤销共轭认证。与认证性能排序选择策略相比,小协方差策略指导的方向选择可以更有效降低融合分数的方差,获得更高的共轭认证性能。

It is imperative to develop bimodal verification fusing palmprint and palmvein,which has high verification accuracy,good practicability and strong anti-spoof ability. 2D-PalmHash code （2DPHC）, as a novel cancelable palmprint verification coding scheme, overcomes the secure vulnerahilities and privacy problems of direct usage of original palmprint features. The framework of 2DPHC can be generalized to cancelable palmvein. Cancelable palm-print-vein verification can be achieved by fusing 2DPHCs of palmprint and palmvein. In order to prevent the increase of computational cost and template data size in bimodal verification, it is ＇necessary to effectively select the orientations, along which the palmprint and palmvein templates are generated. In each model biometric,2DPHCs are generated along four orientations, and the covariances hetween the matching scores of the four 2DPHCs are calculated. According to ＂small covariance＂ selection rule,two 2DPHCs with minimum covariance are respectively selected from palmprint and palmvein as cancelable conjugate templates. The matching scores of the templates are fused with mean rule for cancelable conjugate verification. Compared with the rule of ＂verification accuracy sequencing＂ , the ＂small covariance＂ rule for orientation selection can suppress the variance of the fused matching score more effectively,and accordingly achieve higher conjugate verification accuracy.