Revocable Finger Vein Template Protection Algorithm Based on Pairwise Inner Product Hashing

Abstract

In the existing two-factor revocable biometric template protection scheme, the uniform random permutation hashing (URP-IoM) algorithm is reliable and irreversible. However, this method calculates the Hadamard inner product by randomly permuting the original feature vector, and records the maximum index value, which cannot completely retain all the information of the original biological feature vector. Therefore, this paper proposes a new hashing algorithm based on pairwise inner product (PIP Hashing). Firstly, according to the random permutation seeds generated multiple times, the original feature vectors are randomly arranged multiple times, in order to enhance the irreversibility; Secondly, calculate the Hadamard inner product of the random permutation feature vectors obtained in pairs, and splice them together to better retain the features of the original vector; Finally, randomly select the inner product value vector with the same dimension as the original feature vector from the stitched Hadamard inner product vectors, and then obtain the maximum index from the fixed window. The experimental results show that the PIP Hashing algorithm improves the recognition rate of the algorithm on PolyU and SDUMLA-FV finger vein databases, and meets the revocability and irreversibility standards of revocable biological templates.