Secure latent fingerprint storage and self-recovered reconstruction using POB number system

Abstract

Latent fingerprints are the unintentionally deposited fingerprint impressions gathered from the crime scenes. Many criminal investigation agencies consider latent fingerprints as a significant court accepted evidence. A typical latent fingerprint comes in low quality. Hence, a slight modification in the latent fingerprint may induce a marked shift in the recognition performance. Due to this, wrongdoers behind the crime scenes may try to remove or alter the latent fingerprint information by accessing the fingerprint database. Unlike regular fingerprint enrollment, retaking a latent fingerprint is not always possible. Preserving the latent fingerprints in a single database makes it vulnerable to single-point attacks. Hence, this paper presents a secure way to store and retrieve latent fingerprint information using POB-based (n,n) VSS technique. The proposed method encrypts each latent fingerprint as n secret shares, and stores them in n distinct databases. The distributed storage protects the data from single-point attacks. Along with secure storage, we also introduce a self-recovery mechanism in the case of fingerprint share tampering. The self-recovery mechanism protects the latent fingerprint from different tampering attacks. The proposed method has been evaluated using NIST Special Database4 (NIST-SD4) and IIIT Delhi latent fingerprint datasets. The experimental results show that the proposed technique offers secure distributed storage with lossless reconstruction of latent fingerprint images whenever needed. The proposed self-recovery mechanism enables the recovery of latent fingerprint images even in the case of share tampering.