Convert the image pixels to binary data using key

I am a student learning about Digital Watermarking. When i am reading this article: https://www.sciencedirect.com/science/article/pii/S1319157820303384?via%3Dihub

At the data embedding process, It's said that you can "Convert the watermark image pixels D(x,y,z) to binary data bi after shuffling using the key ‘K’." and later you can recover image from that "binary data". I am not trully understand this, first is how can i convert color image to binary image (maybe use rgb2gray() ?); second is how can i recover from binary to color image using key K (it's impossible)?

Anyone know any image processing method or algorithm or anything can do that?