Photographers use watermarks to protect copyrighted images. Researchers at Google recently created an AI algorithm that can remove watermark from photos in no time.
Google decided to disclose this vulnerability in order to help photographers and stock image communities to better protect copyrighted images.
Image watermarking issues
The watermarking methods that photographers use are very consistent. This is actually their Achilles’ heel. Special algorithms can break this consistency and use it to invert the watermarking process.
More specifically, AI algorithms identify and analyze the watermark image and its opacity, separating it from the rest of the photo. Using watermarked image collections as input, the system can then identify the watermarking patters. After determining which watermarking structure is repeated, the algorithm can then automatically remove watermarks from a new collection of photos.
In other words, if users embed a similar watermark in a collection of images, the watermark then becomes the signal and the images become the noise. This allows image operations to identify the watermark pattern.
Here are the conclusions of this research:
We revealed a loophole in the way visible watermarks are used, which allows to automatically remove them and recover the original images with high accuracy. The attack exploits the coherency of the watermark across many images, and is not limited by the watermark’s complexity or its position in the images. We further studied and evaluated whether adding small random variations in geometry/opacity to the watermark can help prevent such an attack. We found the attack is most affected by geometric variations, which can provide an effective improvement in watermark security compared to current, traditional watermarking schemes.
For more information on how Google’s algorithm identifies and removes watermarking patters, check out the video below:
Using third-party software to remove watermark from photos is a dishonest practice. Copyrighted images should be used only with their owners’ permission.Follow The AI Center on social media:
Join me as I track the latest progress in AI research.
Latest posts by Mary Blaut (see all)
- AI powers staff-free stores: Shop without any human interaction - January 18, 2018
- AI helps researchers improve gene editing results - January 13, 2018
- True Emoji is an AI app that uses your expressions to create animated emojis - November 21, 2017