Today I read a paper titled “Digital watermarking : An approach based on Hilbert transform”
The abstract is:
Most of the well known algorithms for watermarking of digital images involve transformation of the image data to Fourier or singular vector space.
In this paper, we introduce watermarking in Hilbert transform domain for digital media.
Generally, if the image is a matrix of order $m$ by $n$, then the transformed space is also an image of the same order.
However, with Hilbert transforms, the transformed space is of order $2m$ by $2n$.
This allows for more latitude in storing the watermark in the host image.
Based on this idea, we propose an algorithm for embedding and extracting watermark in a host image and analytically obtain a parameter related to this procedure.
Using extensive simulations, we show that the algorithm performs well even if the host image is corrupted by various attacks.