Re: Similarity between integers in a list
Re: Similarity between integers in a list
- Subject: Re: Similarity between integers in a list
- From: Martin Orpen <email@hidden>
- Date: Wed, 19 May 2004 12:27:05 +0100
[originally sent to the OP instead of the list by mistake]
on 19/5/04 3:04 am, BJ Terry at email@hidden wrote:
>
I don't know how well this works in the real world, or how fast you
>
need to be able to do this, but I strongly suspect that Photoshop would
>
be the limiting factor in terms of performance. This script just takes
>
the average difference between all the pixels and compares them to a
>
set threshold. Adjust it for sensitivity. I suspect that 2 is two low,
>
since the total average difference in the first case was 27, and only
>
1.7 for the second set of points, but it depends, as always, on your
>
source images. Additionally, something that humans would consider
>
neutrally colored, like sepia, would probably not be accurately
>
classified by this program, but I'm sure you're familiar with issues
>
such as that, working with Photoshop and all. It's dirt simple though.
>
>
property imageThreshold : 2
>
set image1 to {{71, 94, 97}, {95, 96, 98}, {60, 68, 73}, {41, 49, 56},
>
{21, 23, 36}, {44, 47, 52}}
>
set image2 to {{89, 89, 89}, {82, 82, 82}, {77, 77, 77}, {48, 49, 49},
>
{48, 49, 50}, {66, 68, 68}}
>
>
on neutralp(pixelList)
>
set totalDiff to 0
>
repeat with x from 1 to count pixelList
>
set currentPixel to item x of pixelList
>
set diff to abs ((item 1 of currentPixel) - (item 2 of currentPixel))
>
set diff to diff + (abs ((item 1 of currentPixel) - (item 3 of
>
currentPixel)))
>
set diff to diff + (abs ((item 2 of currentPixel) - (item 3 of
>
currentPixel)))
>
set totalDiff to totalDiff + diff
>
end repeat
>
return (totalDiff / (count pixelList)) < imageThreshold
>
end neutralp
>
>
neutralp(image1)
Thanks for that BJ - I was able to drop that straight in to my script and
have some fun testing the thresholds with a number of images.
A threshold of 2 is great if the client has sent images that have been
"neutralized" in PS. A threshold of 6 would differentiate between images
that had a colour bias, but are still good candidates for a different
separation method.
I've got the threshold set at 10 at the moment and don't seem to be getting
any false positives.
Cheers
--
Martin Orpen
Idea Digital Imaging Ltd -- The Image Specialists
http://www.idea-digital.com
_______________________________________________
applescript-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/applescript-users
Do not post admin requests to the list. They will be ignored.