plowry at scotia-group.com
Mon Oct 30 10:03:48 EST 2000
That was a quick response! Thanks.
Yes it is just summation. The values that are summed are the values of the
splats where they are intersected by the sampling grid. If the input data
are not coincident and their splats overlap at the sampling point then each
splat will contribute a value between zero and the peak splat value (set by
the scale factor). The sum of these will then be the kernel density at the
Thanks for your help. I have references if you would like.
The Scotia Group
At 09:25 AM 10/30/2000 -0400, you wrote:
>I don't see why we couldn't put in some machinery to control the accumulation
>function in vtkGaussianSplatter. Is the function a simple summation? If
>so we can code it pretty quickly.
>At 03:45 PM 10/30/00 +0400, Patrick Lowry wrote:
>>I am using the vtkGaussianSplat routine. It appears to me that it is an
>>application of kernel density mapping. In kernel density mapping, the
>>values of all of the kernels intersected a sampling point are summed. The
>>value at each sampling point can exceed the maximum value of individual
>>kernels. The importance of this is that the summed kernel values can be
>>contoured and accurately show data density (better so than a histogram
>>In the GaussianSplat routine it appears that the splats (kernels) just
>>merge. Even with dense data, the maximum value attained in the sampling
>>grid is that of an individual splat. The GaussianSplat routine has great
>>potential but if it does not allow the characterization of the dense
>>portions of a data set, its use is limited.
>>Is it possible to instruct the GaussianSplat routine sum the splat values at
>>each sampling point? I would appreciate any assistance or guidance in
>>getting true kernel density mapping utility out of the vtkGaussianSplat
>>The Scotia Group, Inc.
>>This is the private VTK discussion list.
>>Please keep messages on-topic. Check the FAQ at:
>>Follow this link to subscribe/unsubscribe:
>This is the private VTK discussion list.
>Please keep messages on-topic. Check the FAQ at:
>Follow this link to subscribe/unsubscribe:
More information about the vtkusers