Secure Visual Image Cryptography: A Password-Authenticated, Share-Based Approach for Confidential Image Protection

Abstract

In this new digital age, protecting sensitive visual information, from identity documents and medical scans to private photographs, is becoming a necessity. However, conventional image-encryption schemes based on ciphertext rely on cryptographic keys that may be lost or intercepted, such that the secured data will be at direct risk in the event of loss. In this paper, we present a Secure Visual Image Cryptography system that removes the ever-elusive single-point key dependency by splitting a secret image into three noise-like shares utilizing randomization and bitwise XOR operations, where no individual share leaks data on the original image. The three shares can only be losslessly combined together with the correct password, which is checked empirically through hash matching. The system, implemented in Python using the Pillow imaging library and hashlib for authentication, is exposed via a Flask-based web interface with separate Admin and User modules designed specifically for share generation, reconstruction, user management, and query handling. Sample images experimentally validate that the shares generated are visually indistinguishable from random noise, that reconstruction is only possible with the complete set of shares and the correct password, and verification shows that each recovered image is pixel-by-pixel identical to its source image. The experimental outcomes reveal that the suggested framework offers a more viable and key-management-free solution for traditional image encryption aimed at secure communication, protected document interchange as well as safeguarding digital assets.