Finger Vein Template Encryption Scheme Based on BioHashing

This paper proposes a cancellable finger vein authentication and template protection scheme based on BioHashing. BioHashing uses a pseudo-random matrix to map the original feature vector to the low-dimensional space, and this projection is a one-way irreversible transformation, which ensures the security of the templates. Aiming at the serious degradation of original BioHashing authentication performance caused by user's token leakage, a new scheme of discretization was proposed to improve the original algorithm. The main idea is to obtain the fixed length coding of the finger vein feature by constructing a binary tree about the projection feature. During the authentication stage, when calculating the Hamming distance, the weight of the coding generated by each layer of the binary tree was set adaptively. Through experiments, the improved scheme can protect the sensitive data of the finger vein templates and improve the recognition performance of the system effectively.