LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

Plot 3D surface for point cloud? 3D convex hull?

Challenge:

 

In my application I end up with small clusters of 3D point clouds representing small objects of arbitrary shape. I have been able to use an implementation of the DBSCAN algorithm to seperate my little point cloud blobs into distinct groupings.

 

Now I would like to draw a 3D shell around these groupings so they can be displayed clearly on a 3D plot, with their shape obvious. 

 

I am able to use the native LabVIEW 3D picture functions to draw plot the point clouds, but I cannot see a good tool within LabVIEW to create a complete 3D surface. A 3D convex hull feature would also be useful in this application. 

 

Something liek this would be nice:

 

http://stackoverflow.com/questions/5492806/plotting-a-surface-from-a-set-of-interior-3d-scatter-poin...

 

In the example code I have attached I can switch between drawing points, lines and triangles but that's not good enough to draw the outline of the 3D shape by itself.

 

I have tried to look at PCL and VTK but it looks like implementation of those in LabVIEW would be non-trivial. I can also use MATLAB, but the processing time associated with the brute force methods I have found so far is pretty long.

 

Note: my point cloud represents the full 3D volume of the object, not just the surface. DBSCAN can return whether a point is in the the middle of the object or on the edge so this feature could be used to feed into logic for drawing a polygonal surface.

 

 

 

 

 

points.png

0 Kudos
Message 1 of 2
(3,665 Views)

Well, LabVIEW has at least two Convex Hull routines.  One, "Convex Hull", is in the Computation Geometry sub-Palette (look inside the Mathematics Palette).  This seems to be a 2D routine, as it has X and Y inputs, but maybe you can cleverly "extend" it to 3D.  Another Convex Hull routine is in the Vision (IMAQ) Palette, designed for gathering a collection of Pixels into an Object.  However, this is also (almost surely) a 2D routine.

 

You probably already know about these, but if not, they may give you some hints and places to start.

 

Bob Schor

0 Kudos
Message 2 of 2
(3,631 Views)