Building GP from point clouds

[Carley Stewart]

Hi,

I'm trying to construct a GP from a series of point clouds which are already in correspondence. Is this possible with scalismo? 

Thanks,

Carley

[Marcel Luethi]

Hi Carley

Yes, this is possible. You essentially follow the steps described in Tutorial 6 ( https://scalismo.org/docs/Tutorials/tutorial06 ), but replace the type TriangleMesh with UnstructuredPointsDomain.

This would look somehow like this:

  // the files with poitn clouds in a standard mesh format (e.g. stl)

  val pcFiles : Seq[java.io.File] = ???

  // read files as mesh and extract point set

  val pointClouds = for (pcFile <- pcFiles) yield MeshIO.readMesh(pcFile).get.pointSet

  // wrap pointSet in a Domain

  val pointDomains = for (pointCloud <- pointClouds) yield UnstructuredPointsDomain3D(pointCloud)

   

  // Build model

  val reference = pointDomains.head

  val dc = DataCollection.fromUnstructuredPointsDomainSequence(reference, pointDomains)

  val pdm : PointDistributionModel[_3D, UnstructuredPointsDomain] = PointDistributionModel.createUsingPCA(dc)

The underlying GP can be extracted from the pdm, using pdm.gp

Best regards,

Marcel

--
You received this message because you are subscribed to the Google Groups "scalismo" group.
To unsubscribe from this group and stop receiving emails from it, send an email to scalismo+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/scalismo/83e05fa3-c03e-4798-a6f2-ba00f3993341n%40googlegroups.com.

[Carley Stewart]

I'm having trouble converting my point cloud data (X,Y,Z coordinates in a text file) into an stl file. Is there a method to create a pointSet straight from a .txt file?

Thanks! 

[Marcel Luethi]

Hi Carley

You can create an UnstructuredPointsDomain directly from a sequence of points as follows:

        val pointSeq = IndexedSeq(Point3D(0,0,0), Point3D(1,1,1))

        val usDomain  = UnstructuredPointsDomain3D(pointSeq)

Hence, you can read the coordinates of the points from a txt file, create a sequence of points and use this sequence in turn to define an UnstructuredPointsDomain.

With that you are at the point where you can use the code I sent in the previously to build a point distribution model.

Best regards,

Marcel

On Mon, Mar 20, 2023 at 4:02 PM Carley Stewart <cgst...@gmail.com> wrote:

--

[Marcel Luethi]

Dear Carley

Great to hear that your project is advancing.

Support for visualizing point distribution models has only been added very recently. You need to upgrade to version 0.92-RC1. Once you have the new version, a simple show should work. You don't need to manually add the argument ShowInScene.

Best regards,

Marcel

On Fri, Mar 24, 2023 at 10:28 AM Marcel Luethi <marcel...@gmail.com> wrote:

Dear Carley

Great to hear that your project is advancing.

Support for visualizing point distribution models has only been added very recently. You need to upgrade to version 0.92-SNAPSHOT. Once you have the new version, a simple show should work. You don't need to manually add the argument ShowInScene.

I hope to be able to release a first release candidate of 0.92 later today or on Monday.

Best regards,

Marcel

On Thu, Mar 23, 2023 at 8:54 PM Carley Stewart <cgst...@gmail.com> wrote:

Super! I feel like I am almost there. Thank you for the help so far. I am able to read and point clouds and display them, and create a model and display samples or the mean, but I cannot display the model. Can you do this with a point cloud? Below is my code so far: 

[Carley Stewart]

Hi Marcel, 

I have been able to upgrade and display the model as a point cloud. Could you explain what the coefficients slider bars are? Do they represent different components of the PCA (slider bar 1 = priniciple component 1) ?

Thanks!

Carley