[FFmpeg-user] find repeated fragments in a video file

[Frans Meulenbroeks]

Hi,

I'm looking for a way to detect repeated fragments in a video file.
About 8 years ago I wrote some code to do this (based upon perceptual
hashing), but that code is owned by my old employer and I have no access to
it any more.
And of course now I need something similar

What I did was roughly: calculate a 32 bit hash covering 5 seconds of
video, and do that for every second of video (or maybe every I-frame).
So an hour of video would give me 3600 hashes. These then could be
processed to find near duplicates (using Hamming distance)
(actually it was a bit more advanced but you get the idea)

Of course I can re-create this code, but I was wondering if there is by any
chance something in ffmpeg that I could use for this purpose (e.g. based
upon mpeg-7 signatures).
Is there some support to do this (e.g. calculating a reasonable good and
compact hash)?

Any suggestion is greatly appreciated!

Frans