[Insight-users] Re: problem with centered affine transform

Ashish Poddar ahpoddar at gmail.com
Fri Mar 25 21:16:29 EST 2005


replying myself...

i found the solution which was basically hidden in the type of
images.. I was using imageJ to read the image, which by default takes
the 8-bit images as unisgned int, and doesnt have an option for signed
8-bit images... however in ITK I was producing the 8-bit signed images
in case of registration (using pixeltype as char) and when manually
rotating the images i was using 8-bit unsigned (using pixeltype as
unsigned char)

putting unsigned char everywhere solved the problem !!

Ashish.


On Fri, 18 Mar 2005 01:18:07 -0500, Ashish Poddar <ahpoddar at gmail.com> wrote:
> the original mail bounced back.. so am attaching the files again in a
> zip !! hopefully it will go thru this time...
> 
> regards,
> Ashish.
> 
> 
> On Fri, 18 Mar 2005 01:12:30 -0500, Ashish Poddar <ahpoddar at gmail.com> wrote:
> > Hi,
> >
> > I am performing a centered affine transformation using the
> > registration loop provided by the ITK Framework. I use a binary filter
> > initially to find out the transform parameters, and then apply those
> > transform parameters using BSplineInterpolation to the original 8-bit
> > image.
> >
> > I have attached 4 images with the mail with the description as under:
> >
> > Fixed Image => the image i used as a fixed image which was basically
> > the MicroCT output
> >
> > Moving Image => the moving image used in the registration loop which
> > was basically the Confocal Laser Microscopy image
> >
> > output After registration => the image i got as the final output of
> > the registration process which was basically after applying the
> > calculated transform parameters to the original moving image.
> > [0.518344, 0.851203, 0.000686798, -0.797247, 0.65689, 0.00598433,
> > -0.0110877, -0.0147881, 0.999488, 99.2418, 102.206, 4.99588, 34.3667,
> > 31.4991, 9.40764]
> >
> > output with manual transform => the image i got by applying manual
> > transforms i.e. rotating the image by 300 degrees with center of
> > rotation same as that of registration output.
> > [0.377265, 0.648206, 0, -0.648206, 0.377265, 0, 0, 0, 0.75, 99.2418,
> > 102.206, 4.99588, -200, 31.4991, 9.40764]
> >
> > I have two questions in this case which I am unable to resolve...
> >
> > 1) i have to apply a translation of X=-200 to obtain the image in the
> > similar location though the registration output says it to be 34.3667
> > which it shows currectly..
> > 2) there are some abnormal brightness spotted found in the image which
> > is the output of registration loop (Image 3) which is not there in
> > output with manual transform
> >
> > If anyone can help me with these I wud appreciate it..
> > If there is any further information that is required in order to
> > resolve this problem, I will be happy to share it..
> >
> > regards,
> > Ashish.
> >
> > --
> > Ashish Poddar
> > Have an acceptable reason for accepting anything.
> > Y:ashish_poddar | MSN:ashish_poddar at yahoo.com
> >
> 
> --
> Ashish Poddar
> Have an acceptable reason for accepting anything.
> Y:ashish_poddar | MSN:ashish_poddar at yahoo.com
> 
> 
> 


-- 
Ashish Poddar
Have an acceptable reason for accepting anything.
Y:ashish_poddar | MSN:ashish_poddar at yahoo.com


More information about the Insight-users mailing list