[Insight-users] Geospatial registration

[Luis Ibanez]

Hi Garrett,

Yes, using a custom interpolator is the way to go.
You can create your own interpolator and simply plug it in the
registration methods.

You probably don't want to use the Conjugate Gradient methods
unless you can verify that your metric is extremely smooth and
monotonic.  Unfortunately, most image metrics tend to be noisy
and that precludes the use of optimizers that are based on
second order derivatives, like the Conjugate Gradient.

Are you registering two images that cover a similar extent ?
Like two images of the same geographical region but taken at
different channels ?
or are you using the registration in order to compose a mosaic
out of images that have a small overlap.

Satellite imagery and Microscopy share a number of similar
needs related to the composition of mosaics from images of
limited field of view.

You may want to look closer at the size and location of the
image regions that you want to be taken into account while
the image metric is computed.  In particular that will
affect your selection of the FixedImageRegion() and eventually
the use of an Image Mask.



    Regards,


      Luis




---------------------
Garrett Potts wrote:

> Hello All:
> 
> 
>> I have adapted our image processing chain by using your import image 
>> filter that takes a raw buffer and adapts it into your processing 
>> chain.   Our transformation parameters are normalized between -1 and 1 
>> and act as multipliers on the sensor's adjustable sigma values.  I 
>> have adapted the Transform to use our image to view transformation 
>> mechanism and map the parameters you send to our internal parameters.  
>>  What I am now trying to do is not let the Registration code do its 
>> own transformation but instead set the floating image manually after 
>> each parameter adjustment then let it do the image comparison to the 
>> fixed image.  Is there a way to override the current registration's 
>> internal resampling of the pixels?  If so this will allow me to adapt 
>> the output of my resampling process and feed it directly to the 
>> registration engine on every parameter adjustment.
>>
> 
> I got this working by overriding my own interpolation function.  I guess 
> this was the way to go?    I am now trying out various optimizers to 
> adjust my parameters.  I am having a tough time matching the floating 
> image from the fixed.  I would like to try the conjugate gradient 
> technique but keep getting the ARMIJO condition unable to be met.  Has 
> anyone tried registration on Satellite imagery?  If so, is there any 
> suggestions you can give for the best methods to use?
> 
> Thank you
> 
> 
> Take care
> 
> Garrett Potts
> 
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 
>