[vtkusers] Find closest point

[nuno.jf]

I vtkCleanPolyData doesn't do what I need. vtkCleanPolyData down-samples my
entire set of points, merging duplicate points, removing unused points and
so on... The problem is that the noise (circled) remains, and the proportion
between the noisy points and the wanted points is maintained, as one can see
bellow.

http://vtk.1045678.n5.nabble.com/file/n4818404/cleanpolydata.png 

I believe the filter available in PCL does exactly what its needed in my
case. I think it should not be very hard to implement, as I only need to,
for each point P, calculate the mean distance from it to some k nearest
points. I then get a medium distance and I can calculate the standard
deviation. I then traverse all the P points again and I remove those whose
mean distance is greater than one standard deviation. (I think... ) My major
question is that, if the filter calculates mean distances for each point P
it can take too long to run. My biggest point cloud has 123k points. I think
I can select only a set of points inside my entire dataset, like the
landmarks in ICP. But I don't know how to select points that represent each
important region in my dataset. Any idea how I can do it?
Nuno








--
View this message in context: http://vtk.1045678.n5.nabble.com/Find-closest-point-tp4816065p4818404.html
Sent from the VTK - Users mailing list archive at Nabble.com.