A quantization-based image authentication system

In this paper, we introduce a robust and secure digital signature scheme for image content authentication, by integrating content feature extraction, watermarking, cryptographic hashing, and digital signature into a unified framework. Given an image, a cryptographic signature is robustly generated based on extracted invariant features and then embedded into the DCT coefficients as watermarks. Quantization is applied to both robust feature extraction and watermarking. Simulation results further demonstrate the practicability of our method in terms of system robustness and system security.