InvZW: Invariant Feature Learning via Noise-Adversarial Training for Robust Image Zero-Watermarking

2025-06-25

AI Summary

To address the insufficient robustness of image zero-watermarks against distortions, this paper proposes an end-to-end deep framework based on distortion-invariant feature learning. Methodologically, it jointly optimizes noise-adversarial training and reconstruction constraints to learn deep features that are both robust to common distortions (e.g., compression, filtering, additive noise) and semantically rich; introduces a trainable reference-code projection module to map features into stable binary watermarks; and employs an adversarial discriminator to enforce feature invariance. The key contributions are: (i) the first unified modeling of distortion invariance and semantic expressiveness in zero-watermarking; and (ii) a multi-bit, fully differentiable, end-to-end trainable zero-watermark architecture. Extensive experiments on multiple benchmark datasets and under complex distortion scenarios demonstrate significant improvements in watermark recovery rate and feature stability over existing self-supervised and deep watermarking methods, achieving state-of-the-art performance.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

This paper introduces a novel deep learning framework for robust image zero-watermarking based on distortion-invariant feature learning. As a zero-watermarking scheme, our method leaves the original image unaltered and learns a reference signature through optimization in the feature space. The proposed framework consists of two key modules. In the first module, a feature extractor is trained via noise-adversarial learning to generate representations that are both invariant to distortions and semantically expressive. This is achieved by combining adversarial supervision against a distortion discriminator and a reconstruction constraint to retain image content. In the second module, we design a learning-based multibit zero-watermarking scheme where the trained invariant features are projected onto a set of trainable reference codes optimized to match a target binary message. Extensive experiments on diverse image datasets and a wide range of distortions show that our method achieves state-of-the-art robustness in both feature stability and watermark recovery. Comparative evaluations against existing self-supervised and deep watermarking techniques further highlight the superiority of our framework in generalization and robustness.

Problem

• Robust image zero-watermarking with unaltered original images
• Distortion-invariant feature learning via noise-adversarial training
• Learning-based multibit watermarking for enhanced feature stability

Innovation

• Noise-adversarial training for invariant features
• Unchanged original image with reference signature
• Learning-based multibit zero-watermarking scheme