In this case image frequency can be useful in many things from image/video compression to hashing images (to efficiently compare if images are similar, even in the presence of cropping etc). So I'm curious as to what in particular motivated posting this now?
I think it's interesting to the modal HN viewer because although FFTs are pretty neat intrinsically, but most people are exposed to FFTs through sound processing (if they are exposed to Fourier Transforms at all, I don't think they were a required part of the CS curriculum at my university, they were only required for the EE courses), so the application of FFTs to images seems like total magic.
The explanation that jpegs use FFTs is ho-hum if you spend a bunch of time doing signal processing at your day job, but there are many intellectually curious developers who spend their day job gluing together CRUD apps.
A lot of the time it's because the link was relevant to a discussion in another thread, and someone pulled it out of that discussion to submit at the top level.