Watermarking is a practice that has been employed since the 13th century, but in digital steganographic practice, only began to be demonstrably adopted in the early 1990’s. In our case, we apply a discrete Fourier transform to convert an image into its frequency components. We then embed an artificial message inside the image (modifying its frequency from random noise into modified non-uniform noise). The goal with this particular notion is not expecting the complete recovery of the embedded message per se, but rather, to distinguish if the image was algorithmically modified at all from random noise after being shared through numerous platforms and modification methods. A better example can be shown visually below:

Last updated