[vorbis-dev] Tarkin at last

Steve Underwood steveu at coppice.org
Fri Jan 5 08:57:13 PST 2001

Lourens Veen wrote:

> > something I'ld like to call an adaptive wavelet, but really isn't a wavelet at all,
> >
> > except for falling into the predict&correct category, based
> > on the concept of edge-directed interpolation
> >     http://www.ee.princeton.edu/~lixin  (download the thesis)
> >
> [snip]
> >
> > Erik.
> Having read part of it and looked at the images, I think that whatever
> we do we need to pay special attention to edges, since accurate edge
> representation helps the perceived image quality a lot (even if the PSNR
> is the same).

If you think about how human vision works you should realise it is only the edges that
matter at all. The visual vortex evaluates LPF'ed differences, and passes these to the
brain. That means if the eye observes a static scene, nothing is sent to the brain at
all. To offset this problem, the servo control of the eyes keeps them constantly
wobbling, so that edges keep generating differences. Seems real clunky to use a
mechanical means to simplify the greyware signal processing, but thats the basic
technique mummy nature same up with.

Its easy to observe this in action. Just saw the top off a monkey, and put its head in a
vice to keep it still. Then you just probe around in its visual cortex while putting a
variety of patterns, in different lighting, in front of it. If you want to see the
effect of the mechanical wobbling, just drug the eye control muscles. That's what they
did at my old college. Pretty gruesome, huh?


--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.

More information about the Vorbis-dev mailing list