Author Topic: Aliasing detection and correction  (Read 5979 times)

a1ex

  • Administrator
  • Hero Member
  • *****
  • Posts: 12564
Aliasing detection and correction
« on: July 19, 2013, 02:05:04 PM »
http://www.math-info.univ-paris5.fr/~moisan/papers/2010-10r.pdf

Who's going to decipher and implement this?

DTSET123

  • Member
  • ***
  • Posts: 187
Re: Aliasing detection and correction
« Reply #1 on: July 19, 2013, 02:18:20 PM »
You?  :P

a1ex

  • Administrator
  • Hero Member
  • *****
  • Posts: 12564
Re: Aliasing detection and correction
« Reply #2 on: July 19, 2013, 02:21:05 PM »
It's way too complicated for me.

mvejerslev

  • Member
  • ***
  • Posts: 103
Re: Aliasing detection and correction
« Reply #3 on: July 19, 2013, 03:28:31 PM »
Ah, the Fourier domain.
5D Mark II, PC

stevefal

  • Contributor
  • Hero Member
  • *****
  • Posts: 528
Re: Aliasing detection and correction
« Reply #4 on: July 19, 2013, 03:48:25 PM »
Maybe the authors can provide a practical recipe. Both are on LinkedIn.
Steve Falcon

bumkicho

  • Member
  • ***
  • Posts: 102
Re: Aliasing detection and correction
« Reply #5 on: July 19, 2013, 03:49:14 PM »
It's way too complicated for me.
WHAT???!!!

Yuppa

  • Freshman
  • **
  • Posts: 61
  • 60D + ML (Nightly)
Re: Aliasing detection and correction
« Reply #6 on: July 19, 2013, 05:30:57 PM »
Not gonna happen, ‘cause:

($1 / 434 + {6 * 2} - (that horseshoe looking thingy / 2) & * 1 + (4550334 * 0) * (1 / .4545 * (12^3)) (uh, looks like a weird letter “E”) - #@# % * 79 <? <> & == 0) * 0

(Note: for symbol replacements, see the “Symbol Replacements Table.”)

See what I mean?
When you care more about capturing DATA, as opposed to WONDERMENT, you've lost your creative SOUL.

AriLG

  • Freshman
  • **
  • Posts: 79
Re: Aliasing detection and correction
« Reply #7 on: July 19, 2013, 06:02:40 PM »
WHAT???!!!
Amazing that he can't do that, eh ?

a1ex, while you're on it... please fix the white balance on these (you should be familiar with white balance by now).

This is the formula : uˆ(ξ) = ˆv(ξ) + ˆv(H(ξ/2)) and g(µ + ξ) + g(µ − ξ).

(in plain English : make a lot of white, and little yellow. yellow is bad).

You got three days.

Thanks

PS
The above is RAW egg  ;D

T3i (main), T2i
------------------
It's not about accuracy,  it's about Aesthetics

ItsMeLenny

  • Hero Member
  • *****
  • Posts: 928
  • 550D
Re: Aliasing detection and correction
« Reply #8 on: July 20, 2013, 04:19:27 AM »
My brother is a maths boy.
He likes this forum because even the people who speak English as their first language don't speak it correctly.
That sentence was an example of itself.

ItsMeLenny

  • Hero Member
  • *****
  • Posts: 928
  • 550D
Re: Aliasing detection and correction
« Reply #9 on: July 20, 2013, 04:38:13 AM »
The article basically does what the wavelet transform does... which is essentially jpeg compression.

Audionut

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 3654
  • Blunt and to the point
Re: Aliasing detection and correction
« Reply #10 on: July 20, 2013, 05:29:07 AM »
What about trying something a little simpler like supersampling.

driftwood

  • Freshman
  • **
  • Posts: 70
Re: Aliasing detection and correction
« Reply #11 on: July 21, 2013, 02:14:24 AM »
Its really tough stuff. Here's a 'good understanding' abstract for budding coders to look at where adaptive and non-adaptive techniques are currently employed.

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.122.2248&rep=rep1&type=pdf
Canon 60D, Canon 5DMK3, Lexar 1000x 128GB CF, Panasonic (shhhh!) GH2s & GH3s. :-)

andyshon

  • Freshman
  • **
  • Posts: 80
Re: Aliasing detection and correction
« Reply #12 on: July 21, 2013, 04:05:39 AM »
Might it help if the demosaicing algorithm was aware of the actual physical layout of the pixels? We're on the edge of my understanding here but I believe for video the 5D3 uses small clusters of pixels, spaced across the chip. If a standard bicubic interpolation was applied to this, pixels from within the cluster would be weighted equally to those which are actually much further away on the chip. If the debayer algorithm could be weighted to compensate for the specific pixel layout perhaps it would suppress some of the aliasing.