Originally Posted by professurreal
Sounds a bit like what is done in video compression where the parts that do not change from frame to frame (e.g. a static background) is redundant information while it is not changing.
Getting rid of this redundant informtation enables the kind of compression rates you get in formats such as divx. At least that's how I understood it to be
Yeah, I got into this topic after looking up about literally 99 different articles related to digital image enhancement / correction / high-fidelity storage functions (both artistic and technical).
I've been studying a little bit of semi-intelligent/sampled image noise reduction software. I've just got a demo version, but it's so powerful, that I learn a lot from it and then I either edit the results or the source material or both.
The nice thing about understanding a little bit of waveform science is that it translates into several different fields simultaneously. So even if you can't comprehend the Mathematics, it starts to make sense if you see enough incarnations of a applied waveform science.
You guys are both helpful in what you both said, too.
I have the intuition that it's also applicable to things like
Adaptive Differential Pulse Code Modulation
or pretty much any type of modern encoding where the input is compared to the output or the metaoutput and both corrected for errors and/or related to a predicted signal. Often it's more efficient to store the difference instead of the value. Or it's more efficient to store whether or not a condition changes rather than merely the exact values of the conditions.
Calculus is deep as heck in this stuff... In a good way!
I didn't really say this stuff right, but the signal to noise ratio of my comprehension is gradually getting better
Also, of course a signal could be sparse if it's like a wire that's picking up intermittent bursts of noise that are low amplitude and or separated by long silences. The long silences can be compressed and low amplitude might help compensate for the complexity of the noise waveform.
IDK, stuff like that.