Privacy-preserving biometric authentication through visual cryptography-based fingerprint protection

Biometric authentication systems demand strong security mechanisms to protect fingerprint data from unauthorized access and leakage. This paper proposes a privacy-preserving framework using a (2,2) Visual Cryptography Scheme (VCS) integrated with Least Significant Bit (LSB) watermarking for secure fingerprint storage and transmission. The input fingerprint image of size approximately 300 × 300 is first converted into binary form and then decomposed into two cryptographic shares, each expanded to 300 × 600 pixels, resulting in a 2× increase in width. Individually, each share appears as random noise and reveals 0% information about the original fingerprint, ensuring information-theoretic security. In the proposed system, one share is stored in the database, while the second share is embedded into an RGB cover image using LSB watermarking. The embedding layer is determined using employee ID mod 3, and additional security is introduced through row-wise random flipping with a probability of 0.5, producing 2^600 possible flipping combinations for a 600-row image. During authentication, both shares are combined using XOR operations to reconstruct the fingerprint, and validation is performed based on a bit mismatch threshold of 10%. Security analysis indicates that an attacker must explore approximately 3 × 2^600 × 2^(300×600) possible combinations, making brute-force attacks computationally infeasible. Experimental results demonstrate low computational complexity, fast decryption, and accurate reconstruction, making the proposed system suitable for real-time biometric authentication and resource-constrained IoT environments.