Load legacy vtk files into the MDWorkspace structure

[Owen Arnold]

This is an exploratory piece of work requested by Jon Taylor, based on the vtk data file fly.vtk, which contains a real-space 3D image of a fly head.

Generate a new algorithm called LoadVTK, which will read a vtk file and output a 3D IMDWorkspace.

[Owen Arnold]

refs #7252. Algorithm template.

Changeset: b3dc737b7469a5f5cda3a6be793cc7ce2b8edbe1

[Owen Arnold]

refs #7252. Read metadata into md histoworkspace.

Changeset: 7353c11df611a4040c3697453ce22508aa9f20ec

[Owen Arnold]

refs #7252. Properties to specify vtkDataArray names

Changeset: 301978b7609b5a995813603a11762c852e7c1b83

[Owen Arnold]

refs #7252. And Import the data too.

Changeset: 1aeffbac24ab417f94a7f1d1f1c9b43a64e0f0ac

[Owen Arnold]

Data was taken from the Paraview examples. Used the paraview python filter to add two scalar arrays of type unsigned int onto the original data. python is as follows:

pdi = self.GetInput()
pdo = self.GetOutput()
newData = vtk.vtkUnsignedShortArray()
newError = vtk.vtkUnsignedShortArray()
newData.SetName("scalar_array")
newError.SetName("error_array")
npoints = pdi.GetNumberOfPoints() 
for i in range(npoints):
  newData.InsertNextValue(10)
  newError.InsertNextValue(1)
pdo.GetPointData().AddArray(newData)
pdo.GetPointData().AddArray(newError)

[Owen Arnold]

refs #7252. Reporting and test updates.

Changeset: b11a430cfede07a05f587a9a8aa669d27c15e9ac

[Owen Arnold]

refs #7252. Merge branch 'feature/7252_load_vtk' into develop

Conflicts:

Code/Mantid/Framework/Algorithms/src/IntegrateByComponent.cpp

Changeset: 9d9593886c9290b3ab339f17905ff40997b7b94a

[Owen Arnold]

refs #7252. Fix OpenMP loop on windows.

Changeset: 084b31394565dfbd83501ae9cae7692348e16d96

[Owen Arnold]

Tester:

You'll need to be located at ISIS to do the following. Load the Fly.vtk dataset currently in the \Olympic\Babylon5\public\mantid directory into Mantid using LoadVTK. Open it in the slice viewer and you should see something similar to the attached image.

[Owen Arnold]

refs #7252. Algorithm template.

Changeset: e6048a8af990e9179622c753cfa32d114b4bdd1a

[Owen Arnold]

refs #7252. Read metadata into md histoworkspace.

Changeset: 39ff44181b83c92be4046784341383037ec5beee

[Owen Arnold]

refs #7252. Properties to specify vtkDataArray names

Changeset: 0c201b2d4e2dad6c891bc3f5651461c8feafc865

[Owen Arnold]

refs #7252. And Import the data too.

Changeset: 427b83a3301db61d831e530e6a0d2a817cf8ac59

[Owen Arnold]

refs #7252. Reporting and test updates.

Changeset: 548bbf1f3ff3b7cb5d6f09cce48ae37fccdbe2f2

[Owen Arnold]

refs #7252. Fix OpenMP loop on windows.

Changeset: c3d195d178100ce5048dc636f01244fcf34fe76a

[Owen Arnold]

refs #7252. Implement thresholding on image.

Changeset: 80f2a10501ce9c4f0210e1948d50e2a970ac89ac

-- This was a mistake. should have gone against 7264

[Martyn Gigg]

Needs a memory check as the current layout of MDHistoWorkspaces requires one contigous array for all of the data and allocating this can kill a machine without enough memory.

[Owen Arnold]

refs #7252. Add predicted size check.

Changeset: a17060aa8b75e3c2d70bca8e0367808e8f0cd9d3

[Martyn Gigg]

I still get all of my memory being used. I think the check is in the wrong place as it is the MDWorkspace constructor that is allocating the large chunk of memory so the check needs to go before that.

There is also a memory leak if the algorithm aborts due to there not being enough memory as the output->Delete() is never called.

[Owen Arnold]

refs #7252. Further fix memory issues.

Changeset: 38b8b4e30b60a127fece9b073409eec7e2f8f764

[Martyn Gigg]

I can now successfully load the iron_protein but get an error about the fly data set as expected.

[Stuart Campbell]

This ticket has been transferred to github issue ​8098