Post

Tim Hutton - 2015-06-16 09:06:34+0000 - Updated: 2015-06-16 09:06:34+0000

The Curious Phenomenon of Stochastic Resonance

Shared with: Public, Michael Prescott

+1'd by: Rubén Perblac, Håkan Holgersson, Violeta Nataniel, Dave Wonnacott, Paul Gray, Roice Nelson, Michael Prescott, Dmitry Shintyakov, Jochen Fromm, Michal Čaplygin

Reshared by: Mirko Bulaja, William Rutiser

Michael Prescott - 2015-06-16 13:47:17+0000

That's super neat. I'm not sure what to make of it! I wondered whether this was a simple trick of some sort, but Wikipedia suggests that stochastic resonance is a very deep phenomenon, useful for recovering weak signals in real situations. (Mind totally blown.)

Tim Hutton - 2015-06-16 13:54:05+0000

+Michael Prescott Yes. It seems too simple to be true. In image processing it's called dithering (as pointed out on HN: https://news.ycombinator.com/item?id=9723443) but nice to know that the same phenomenon crops up all over and has a name.

Paul Gray - 2015-06-16 15:34:42+0000

So by increasing the entropy and with that, the large blobs more broken up, then any filter processing will be less harsh in effect.  If you do this with say a movie with random dots each frame as this then when you compress it certain aspects will also appear sharper as the lossless image compression is unable to abstract as much down into the smaller components, ie shapes that move when those shapes now have random dots/changes to them each frame.

Another consideration is that what we see is in vastly more detail than we need to identify things and it is somewhat image overloaded and by remove some of that you make it easier to see things.  But as we have relational memories then once we have seen an original we can recapture that detail in lesser versions later on.  Some of Andy Warhols work makes use of that and the classic Monroe picture being one of the best examples.

Still, pointillists will go, so what :).

This post was originally on Google+