I'm doing some research on digital watermarking and came across an article that embeds QR codes on mostly grayscale images and the extraction part for color images is quite confusing the article states:
In the simplest cases, when the watermark was located over the relatively uniform area, a modification of histogram curves, that results in changes to contrast and brightness, brings satis- fying watermark readability. In more complex situations, when the watermark was located over the non-uniform elements, such as colorful pictures, the method allowing to find the watermark is histogram-based image segmentation.
They said that extraction part was done using GIMP(I've already tried but failed).
I couldn't find anything on histogram image segmentation but found about histogram equalization and threshold methods but none of them were successfull, below is my failed attempt:
The embedding process is on the Y Channel of the original image and in the extraction process I'm reading the watermarked image as jpeg(colored) and converting to YUV.
How can i properly amplify the difference of luminance between pixels of color images?