Enhanced fuzzy vault scheme on palm vein biometric template protection using fast walsh Hadamard transformation

Abstract

The widespread use of biometrics as a means of identification continues to rise a security issue for the user. Once a biometric template is compromised, it is considered broken as biometrics cannot be revoked and thus making it obsolete. This is solved by adapting to the biometric template protection scheme. However, not all approaches can meet all the requirements of the biometric template protection scheme. The first requirement is diversity where a biometric template should not link to any other template that is produced from the same original template. Next is revocability where a compromised template should be able to be replaced by a new template generated from the same original biometric template. The third requirement is irreversibility where the transformed template should not revert to the original template and lastly, the fourth requirement is a performance where the performance of the authentication system should not be affected by any methods to secure the biometric template. A fuzzy vault scheme has been widely used in the biometric template protection approach. However, the fuzzy vault scheme only provides irreversibility and performance. Thus, fast Walsh Hadamard Transformation (FWHT) hybridized with fuzzy vault namely FWHT-FV is proposed to equip fuzzy vault scheme with diversity and revocability. The palm vein dataset from Tonji University is used in this study. Then, pre-processing and feature extraction is applied using the Local Binary Pattern (LBP). The result showed that the proposed FWHT-FV achieved 1.25% FAR, 1.75% FRR, and 5.75% EER. The proposed FWHT-FV has improved the accuracy, sensitivity, and specificity by 12.87%, 16%, and 11.75% respectively. Lastly, security analysis showed that the proposed FWHT-FV is able to equip the fuzzy vault scheme with the diversity and revocability properties of the biometric template protection requirements. Hence, the proposed hybrid approach enables the fuzzy vault scheme to meet all the biometric template protection requirements which are diversity, revocability, irreversibility, and performance as well as benefiting the society and industries in the biometric template authentication field.