gizmodo

03-11-2013, 06:33 AM

I have a data set that is sparse and not organized in any way. I have tried ordering the data based on the X and Z axes. This works, but the surface isn't very useful due to the fact that the data is so sparse it has very sharp peaks and valleys, where I am looking for a nice smooth surface.

The type of data I have is in X, Y, Z format. In my case the ranges for the data along each axis are vastly different. For example the x-axis may be Engine RPM which could go from 1000-8000, where the z-axis could be Pressure Ratio, and be 0.5 - 3. The y-axis, the data I'm interested in plotting, can be any range of number based on what the user selected. What I have decided to do is simply ignore the actual values for the x and z axis, and give them a range from 0 to max and just plot the actual y values. Then I figure I can just draw axes to represent the actual range. That should solve my problem of the chart being unreadable due to differences in range.

I have been doing a lot of reading and keep coming across posts where people mention "Delaunay Triangulation". I've also seen some posts referring to hulls. I have been researching those two concepts, but I'm having trouble understanding how I would go about running the triangulation, what that would output and how I would ultimately render it. I looked into things like MeshLab, Point Cloud Library and various other projects, but those seem to provide far more functionality than I need. What I'm looking for is a way to perform the Delaunay Triangulation on my data, and then display it so the user can pan, tilt and zoom the surface. Any help would be appreciated.

The type of data I have is in X, Y, Z format. In my case the ranges for the data along each axis are vastly different. For example the x-axis may be Engine RPM which could go from 1000-8000, where the z-axis could be Pressure Ratio, and be 0.5 - 3. The y-axis, the data I'm interested in plotting, can be any range of number based on what the user selected. What I have decided to do is simply ignore the actual values for the x and z axis, and give them a range from 0 to max and just plot the actual y values. Then I figure I can just draw axes to represent the actual range. That should solve my problem of the chart being unreadable due to differences in range.

I have been doing a lot of reading and keep coming across posts where people mention "Delaunay Triangulation". I've also seen some posts referring to hulls. I have been researching those two concepts, but I'm having trouble understanding how I would go about running the triangulation, what that would output and how I would ultimately render it. I looked into things like MeshLab, Point Cloud Library and various other projects, but those seem to provide far more functionality than I need. What I'm looking for is a way to perform the Delaunay Triangulation on my data, and then display it so the user can pan, tilt and zoom the surface. Any help would be appreciated.