[Insight-users] Can ITK register successive point cloud data files?

[Luis Ibanez]

Hi Darren,

Yes, ITK can be used for registering two clouds of points.
The typical method for doing this is ICP (Iterative Closest Point).

You will find an implementation of this method in ITK.

In practice you can probably run directly the ICP example
over your data.


Please look at the examples

     Insight/Examples/Registration
               IterativeClosestPoint1.cxx  (2D case)
               IterativeClosestPoint2.cxx  (3D case)
               IterativeClosestPoint3.cxx  (3D with distance map)


You can simply use [IterativeClosestPoint2.cxx] directly.
The input for this example is expected to be in a very
plain file format. You will find input examples under


      Insight/Examples/Data/
                IterativeClosestPointFixedPoints2.txt
                IterativeClosestPointMovingPoints2.txt


It just have the X,Y,Z coordinates of the points in
ascii format.


It is likely that you will have to fine tune the parameters
of the registration in order to make it converge for your
data.

In particular you should pay attention to the translation
scaling (look at the example code for details), and to the
Number of iterations and gradient tolerance. These last
parameters control the convergence criterion that stops the
registration.



    Regards,


       Luis


--------------------
Darren Hill wrote:

> I have just started looking into ITK (and am not a programmer) and need some
> help.  I have been developing a system that uses a laser to accurately scan
> a cylindrical surface.  The laser scans the surface by moving in the
> circumferential direction and then advances axially so that an overlapping
> section is created between successive strips of data.  Each strip is
> approximately 50 mm (2") wide and the circumferential length of the strips
> are constant and are dependant on the diameter of the cylinder.  The
> apparatus consists of a cylindrical track for circumferential motion and a
> linear track for axial motion.  The system that has been constructed thus
> far aligns the strips to within +/- 1 mm (0.040") but is limited by the
> rigidity of the system.
> I am using TecPlot to visualize the date but I would like to know if ITK can
> register 3D point cloud data and if so would someone be interested in
> providing me with a basic software program that would make the minor
> adjustments that I need (translation and rotation) to get the strips of data
> to be accurately registered relative to one another?
> Sample data is available at your request (The 3 dimensional data (X ,Y and Z
> coordinates) are in CSV format).
> Any help or recommendations would be greatly appreciated.
> Regards,
> Darren
> 
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
>