How does Czkawka group images?

[Clouud9]

For context, I've been trying to create a similar app as a personal learning project, only it's in Java.

I've already implemented my own perceptual hash and hamming distance functions, but I don't understand how Czkawka is able to create groups within a specific threshold. I've tried looking over the code, but Rust syntax is quite foreign to me and I have no idea what is going on. Is there anybody who can give me some help?