[Insight-users] arbitrary image registration

Luis Ibanez luis.ibanez at kitware.com
Wed Aug 11 11:37:09 EDT 2004


Hi Yannick,

You will have to guess a reasonable origin for the second image
with respect to the first image. Your guess doesn't have to be
very precise since you are going to do the registration anyways.
However, the better your guess is, the better are your chances
of registering the two images correctly.

As a another option, you could set an arbitrary origin to 0,0
for the second image and do a rigid Landmark-based registration
in order to initialize your registration process. In that context,
you will select features from image A that match features from
image B. In theory 3 points (landmarks) will be enough in 2D.
In practice you probably want to go with 6 or more.

You can use the ICP algorithm for doing this pre-registration, and
use the results for initializing any other type of registration and 
refining the transform.

You will find examples on the use of the ICP algorithm both
for 2D and 3D in

    Insight/Examples/Registration/
                        IterativeClosestPoint1.cxx
                        IterativeClosestPoint2.cxx
                        IterativeClosestPoint3.cxx

Note that ICP may actually be an overkill for this initialization,
since you can compute a closed for solution given that you are
manually associating the Landmarks from image A to the landmarks
of image B.  ICP is more interesting for cases where you don't
have associations available.



BTW: do you have valid values for pixel spacing in both images ?


   Regards,



      Luis



-----------------------
Yannick Allard wrote:
> Hi everyone,
> 
> I have a problem with image registration. I Currently want ot registered two
> remote sensing images. The problem I have is the following :
> 
> I set the origin of the fixed image to [0,0]. But I dont know the origin of the
> moving image in respect of the origin of the first. Since the registration
> algorithms use the origin and spacing, I have a serious problem.... Is there a
> way to solve this problem.
> 
> Thank you in advance
> 
> Yannick
> 
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 






More information about the Insight-users mailing list