CASIAL: Geometric Distortion Robust Image Watermarking

2026-07-29

AI Summary

This work addresses the limited robustness of existing deep learning–based watermarking methods against geometric transformations—such as cropping, rotation, and scaling—which often cause information loss due to region removal or desynchronization. To overcome this, the paper introduces CASIAL, a novel framework that uniquely integrates cover-image-aware global message diffusion with geometrically invariant feature alignment. Specifically, it employs a spatial attention mechanism to distribute watermark information globally across the image and incorporates an invariant alignment learning module to construct geometrically invariant representations. The resulting end-to-end trainable architecture significantly enhances robustness against geometric distortions, substantially outperforming eleven baseline methods across six types of geometric attacks while maintaining high visual quality and strong generalization under both signal and photometric transformations.

Abstract

Deep learning-based watermarking has shown strong robustness against non-geometric distortions, yet its performance under geometric transformations remains limited. Such transformations induce two fundamental failure modes: region removal, such as cropping or masking, which eliminates the information carried by removed pixels, and desynchronization, such as scaling or rotation, which misaligns pixel positions and disrupts decoding. We argue that achieving geometric robustness requires two essential properties: (1) global spread of the watermark message, ensuring resilience even when large regions are removed, and (2) geometry-invariant representations, enabling decoding to remain synchronized despite spatial transformations. Building on these insights, we propose CASIAL, a geometric distortion-robust watermarking framework with cover image-aware message spreading (CAS) strategy and invariance alignment learning (IAL) module. CAS tightly couples watermark bits with cover image features and distributes them adaptively across the entire image, enhancing per-pixel information capacity and robustness to region removal. IAL leverages spatial attention to capture cross-pixel dependencies and align perturbed features into a shared geometry-invariant representation space, mitigating failures due to desynchronization. Across six challenging geometric transformations, CASIAL achieves substantially stronger robustness than eleven prior baselines while preserving high visual quality. It also maintains competitive performance under six signal distortions and four photometric transformations. Notably, although trained only with white-box distortions, CASIAL also exhibits strong transfer robustness to unseen black-box distortions. Comprehensive experiments demonstrate the broad robustness and superior visual quality of our method.

Problem

• geometric distortion
• image watermarking
• region removal
• desynchronization
• robustness

Innovation

• geometric distortion robustness
• cover image-aware spreading
• invariance alignment learning
• deep watermarking
• spatial attention