Quote:
Originally Posted by quantum-x
Maybe I misused the word 'checksum' in my post.
Imagine you divide a frame up into a 3x3 grid, and from each cell take an average hue, brightness, R/G/B channel saturation average, and store that information. Do that for every N frames of a video, and you'll have a unique fingerprint for each video.
When you're processing a video, you could easily make comparisons within a tolerance of X, and you've got the solution there.
|
heh, I wouldn't say making a tolerance based on colors like that would be an easy thing. I suppose that could work but room for error and again quite a bit of resources on both ends scanning every movie like that.