Is it possible to represent 3D surfaces the same way we can represent sound, so that analysing sound is equal to analyzing a specific 3D model?
For example, music is often decomposed into frequencies via Fourier Analysis. These frequencies can then be fingerprinted and compared.
Is such a decomposition (or another) also possible for 3D models? I came across this paper but don't understand it that much yet.
Ideally, this transformation would be rotationally invariant, so that rotating a model yields an equivalent (or similar) decomposition.
Even if they are not fleshed out, I would love to hear your ideas!
Thanks in advance!
