Robust Watermarking Using Generative Priors Against Image Editing: From Benchmarking to Advances
Shilin Lu, Zihan Zhou, Jiayou Lu, Yuanzhi Zhu, Adams Wai-Kin Kong
2024-10-25
Summary
This paper discusses a new method for watermarking images that protects copyright by ensuring that watermarks remain intact even after advanced image editing.
What's the problem?
Current watermarking techniques struggle to maintain their effectiveness when images are edited using advanced methods, such as those powered by AI. This makes it easy for people to remove or distort watermarks, which can lead to copyright violations and challenges in protecting the ownership of images.
What's the solution?
The authors introduce W-Bench, a comprehensive benchmark to test how well different watermarking methods hold up against various image editing techniques. They also propose a new watermarking method called VINE, which improves the robustness of watermarks by analyzing how image editing affects frequency characteristics and using a powerful pre-trained diffusion model to embed watermarks more effectively. This approach ensures that the watermarks are less noticeable while still being difficult to remove during edits.
Why it matters?
This research is important because it enhances the ability to protect copyright in images, which is crucial in a world where digital content is easily shared and manipulated. By developing more resilient watermarking techniques, this work helps artists and content creators safeguard their intellectual property against unauthorized use.
Abstract
Current image watermarking methods are vulnerable to advanced image editing techniques enabled by large-scale text-to-image models. These models can distort embedded watermarks during editing, posing significant challenges to copyright protection. In this work, we introduce W-Bench, the first comprehensive benchmark designed to evaluate the robustness of watermarking methods against a wide range of image editing techniques, including image regeneration, global editing, local editing, and image-to-video generation. Through extensive evaluations of eleven representative watermarking methods against prevalent editing techniques, we demonstrate that most methods fail to detect watermarks after such edits. To address this limitation, we propose VINE, a watermarking method that significantly enhances robustness against various image editing techniques while maintaining high image quality. Our approach involves two key innovations: (1) we analyze the frequency characteristics of image editing and identify that blurring distortions exhibit similar frequency properties, which allows us to use them as surrogate attacks during training to bolster watermark robustness; (2) we leverage a large-scale pretrained diffusion model SDXL-Turbo, adapting it for the watermarking task to achieve more imperceptible and robust watermark embedding. Experimental results show that our method achieves outstanding watermarking performance under various image editing techniques, outperforming existing methods in both image quality and robustness. Code is available at https://github.com/Shilin-LU/VINE.