Watermarks are placed on copyrighted images like stock-photos in order to keep people from using them without permission or without paying. And manually removing them requires Photoshop skills and tim.
However, Google’s research division has revealed just how easy it is for computer algorithms to bypass standard photo watermarking practices, stripping those images of copyright protection and making them vulnerable to reposting across the internet without credit. The research, presented at a leading computer vision conference in Hawaii back in July, is described in detail in a paper titled, “On the Effectiveness of Visible Watermarks.”
“As often done with vulnerabilities discovered in operating systems, applications or protocols, we want to disclose this vulnerability and propose solutions in order to help the photography and stock image communities adapt and better protect its copyrighted content and creations,” Tali Dekel and Michael Rubinstein, Google research scientists, explain in a post published on Google’s research blog.
The trick is to take lots of images — we’re talking hundreds or thousands of photos — with the same watermark and use software to detect repeating structures. With enough examples, the watermark becomes the signal and all of the photos become noise. The watermark pattern can then be removed in totality from the image without reducing the quality of the image itself.
However, along with its method of seamlessly lifting watermarks, Google also provides a way to counteract it. Changing the watermark’s position on the images randomly for each photo doesn’t stop the software from doing its thing, nor does changing how opaque the watermark is. But randomly warping the mark just slightly for each image does prevent the program from removing it in full.