[vtkusers] Interpolation and Rotation of 3d dataset

[David Gobbi]

Reading 16-bit data is easy if the data is raw, the class vtkImageReader2
can read raw data of any type.

Iterating through a volume in VTK is generally done within the
VTK filters, i.e. it is not something that people would usually write
their own code for.  For example, when I do image registration in
VTK, I use custom VTK classes to evaluate the metrics, e.g.
vtkImageSquaredDifferenceMetric, and these are not part of the
main VTK package.  Most people who are doing image registration
choose ITK, because has the image registration infrastructure built-in.

  David

On Wed, Nov 17, 2010 at 4:38 PM, Kevin Neff <kevin.l.neff at gmail.com> wrote:
> David,
>
> I'm using SIH, which is less prone to getting stuck in local minima.  I'm
> trying to align two volumes that differ mainly in radial orientation.  The
> confocal microscope can't image through the entire sample, so we're
> re-constructing from 4 z-stacks that are 90 deg apart.
>
> I'm using a sum of squared intensity differences in the objective function.
> I assume iterating though a volume and getting the attributes of the voxels
> is possible.
>
> Is there an easy way to read in 16-bit unsigned data?
>
> KLN
>
>
>
> On Wed, Nov 17, 2010 at 5:08 PM, David Gobbi <david.gobbi at gmail.com> wrote:
>>
>> Hi Kevin,
>>
>> You can resample and rotate a volume fairly easily in VTK with
>> vtkImageReslice.  You would have to keep your optimization code
>> because the only general-purpose optimizer that VTK has is a
>> Nelder-Mead simplex (vtkAmoebaMinimizer).  The big question is
>> how you would extract whatever parameter/statistics from the images
>> that you need for the optimization -- VTK's image classes are focussed
>> more on visualization than on analysis.
>>
>>  David
>>
>> On Wed, Nov 17, 2010 at 3:54 PM, Kevin Neff <kevin.l.neff at gmail.com>
>> wrote:
>> >
>> > I am evaluating ways to perform a non-interactive interpolation and
>> > rotation
>> > of a 3-d dataset.  Is VTK an appropriate choice for this task?
>> >
>> > I have non-isotropic 3-dimensional data from a confocal microscope.  I
>> > will
>> > have to interpolate to define cubic voxels.  Then, I need to rotate the
>> > volume many times.  The rotations will be directed by a global
>> > minimization
>> > routine.
>> >
>> > I have coded some of this, but it's buggy and I need to figure out if
>> > it's
>> > worth finishing or using another solution, such as VTK.
>> >
>> > KLN
>> >
>> >
>> > ----------
>> > Kevin Neff, Ph.D.
>> > Mayo Clinic College of Medicine
>> > Rochester, MN
>
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at
> http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the VTK FAQ at:
> http://www.vtk.org/Wiki/VTK_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.vtk.org/mailman/listinfo/vtkusers
>
>