How to transform a RGB matrix to a weighted adjacency matrix?

I have the following 604x604 pixel image A

This image A represents connectivity strenght between particular vertices, theoretically this is thus a 47x47 connectivity matrix. The colours represent the connectivity strenght between two vertices as indicated with a scale seen on the image B (which is 456x12 pixel image) below:

Here the absolute left (blue) indicates a connectivity strenght of 0, while the absolute right (red) indicates a connectivity strenght of 5. The goal I have in mind is to eventually output a 47x47 weighted adjacency matrix with entry value that represent the connectivity strengh (between 0 and 5) of each block as seen in image A.

I think this shouldn't be that difficult, yet I can't seem to think of a simple way to do it.

Answer 1

[Steve Eddins]

Borys, I tackled your problem and wrote a blog post about it:

"Transforming a color image to a weighted adjacency matrix"

[Borys Bulka]

Wow thank you so much for your extensive post about it! I am going to study it and apply it myself!