Prev: the Percentage of Energy Loss (PELD)
Next: Stateflow dashed transitions
From: Image Fun on 2 Aug 2010 15:20

> Well a simple brute force method would be:
> 1) Find the indices of the most left point and top most point of each.
> 2) Pad the lower sides of the first image so you have room to float the second image around.
> 3) Based on the indices you found earlier and a global transformation(pad amount); float the window around to minimize the error; which in this case would be the minimum absolute difference.

But I want a code that does this automatically without any manual aid. And the time taken for execution should also be very less. Isn't there any function in MATLLAB that will make this task easier?
From: Sean on 2 Aug 2010 15:40
"Image Fun" <grkkrg(a)rediffmail.com> wrote in message <i375p6$pa4$1(a)fred.mathworks.com>...
>
> > Well a simple brute force method would be:
> > 1) Find the indices of the most left point and top most point of each.
> > 2) Pad the lower sides of the first image so you have room to float the second image around.
> > 3) Based on the indices you found earlier and a global transformation(pad amount); float the window around to minimize the error; which in this case would be the minimum absolute difference.
>
> But I want a code that does this automatically without any manual aid. And the time taken for execution should also be very less. Isn't there any function in MATLLAB that will make this task easier?

What about this isn't completely automatic. Finding the lowest coordinates - automatic, padding - automatic, floating window with error calculation saving the minimum error - automatic. I would imagine this would take, on an image your size, start to finish - about .08 seconds; or faster than you can click a mouse button.

Functions to help you:
max,size,for loop, (if Image Processing TBX) padarray, '-' -> that's a minus sign, <=, abs, sum, (:) ->colon operator to turn matrix into vector...

If you meant you want us to write the function for you that's probably not going to happen. If you want us to help you: show us what you've done and we can point out the bugs or give hints.
From: Image Analyst on 2 Aug 2010 16:48
If you're looking for automated image registration methods, look here:

http://iris.usc.edu/Vision-Notes/bibliography/contentsmatch-pl.html#Registration,%20Matching%20and%20Recognition%20Using%20Points,%20Lines,%20Regions,%20Areas,%20Surfaces

or here:

http://www.mathworks.com/matlabcentral/fileexchange/?term=registration
From: Image Fun on 11 Aug 2010 13:40
The solution you have given works if the two images are of nearly the same size. What if the they aren't? What if the second image has just a 'D'?
My final aim is to get a percentage match on the basis of differing pixels.
First  |  Prev  | 
Pages: 1 2
Prev: the Percentage of Energy Loss (PELD)
Next: Stateflow dashed transitions