SpreadMark: Robust Image Watermarking via Spread-Spectrum Embedding

2026-08-04

AI Summary

This work addresses the challenge that existing image watermarking methods struggle to simultaneously achieve high imperceptibility and robust detection under deliberate removal attacks such as regional perturbations and latent-space sparsification. The authors propose the first integration of classical spread-spectrum techniques into a neural post-hoc watermarking framework, leveraging pseudo-random codewords for bit-level global diffusion. By incorporating cover-aware representation learning, matched-filter recovery, and parallel convolutional decoding pathways—trained under sparsity-aware supervision—the method substantially enhances robustness. Evaluated on COCO and DIV2K datasets, it is the only approach that maintains high detection rates under both regeneration attacks and latent sparsification, while also demonstrating strong resilience to JPEG compression and additive noise, alongside excellent visual imperceptibility.

Abstract

Invisible image watermarks are increasingly used for deepfake detection and provenance tracking, where they must survive not only incidental distortions but also deliberate removal. We revisit spread-spectrum embedding, a classical watermarking principle, inside a modern neural post-hoc watermarking architecture. Our starting point is a measurement: in existing encoder-decoder schemes each message bit occupies only a small fraction of the image, a shared contributing factor to their fragility, since removal then need only disturb the region a bit occupies. SpreadMark instead spreads each bit as a dense pseudo-random codeword over the whole image and recovers it by matched-filtering a learned cover-suppressed chip representation, with a parallel convolutional decoding path and sparsification-aware training. A conditional chip-space analysis shows that, under a codeword-independent perturbation model, dense spreading increases the budget required to disrupt matched-filter recovery. Evaluated on COCO and DIV2K against nine schemes, SpreadMark is the only evaluated method retaining high detection under both the regeneration and the latent-space sparsification settings we test, with competitive JPEG and additive-noise robustness. It keeps the embedded watermark imperceptible, maintaining high perceptual quality on both COCO and DIV2K.

Problem

• image watermarking
• robustness
• spread-spectrum
• deepfake detection
• watermark removal

Innovation

• spread-spectrum watermarking
• dense pseudo-random codeword
• matched-filter recovery
• sparsification-aware training
• cover-suppressed representation