Learning to Immunize Images for Tamper Localization and Self-Recovery

TL;DR

This paper introduces Imuge+, an advanced neural network-based method for protecting images against tampering by enabling auto-recovery and accurate tamper localization, demonstrating superior robustness and performance over existing schemes.

Contribution

Imuge+ employs invertible neural networks and a novel JPEG simulator to enhance image immunization, tamper localization, and self-recovery capabilities.

Findings

Accurate tamper localization in real-world tests

High-fidelity content recovery after tampering

Outperforms state-of-the-art passive forensic schemes

Abstract

Digital images are vulnerable to nefarious tampering attacks such as content addition or removal that severely alter the original meaning. It is somehow like a person without protection that is open to various kinds of viruses. Image immunization (Imuge) is a technology of protecting the images by introducing trivial perturbation, so that the protected images are immune to the viruses in that the tampered contents can be auto-recovered. This paper presents Imuge+, an enhanced scheme for image immunization. By observing the invertible relationship between image immunization and the corresponding self-recovery, we employ an invertible neural network to jointly learn image immunization and recovery respectively in the forward and backward pass. We also introduce an efficient attack layer that involves both malicious tamper and benign image post-processing, where a novel distillation-based JPEG simulator is proposed for improved JPEG robustness. Our method achieves promising results in real-world tests where experiments show accurate tamper localization as well as high-fidelity content recovery. Additionally, we show superior performance on tamper localization compared to state-of-the-art schemes based on passive forensics.