smoothing a surface

[LambdaDigamma]

Hi all Labviewers,

This is voltage I am getting from an array. This is actually a frame of a 3d-surface video. Is there a way I can ''smooth'' the surface? Any ideas welcome. Thank you for your time

[billko]

---

@LambdaDigamma wrote:

Hi all Labviewers,

This is voltage I am getting from an array. This is actually a frame of a 3d-surface video. Is there a way I can ''smooth'' the surface? Any ideas welcome. Thank you for your time

---

All those zeroes to me means a very fine resolution.  Do you need that kind of resolution, or can can you change the scale to something more reasonable?

[altenbach]

There are many ways to smooth a 2D set of data. What kind of resolution to you need to retain?

(Try e.g. a 2D convolution with a gaussian kernel of appropriate size)

(In the most extreme case, you would replace all elements with the average of all elements, but you are probably looking for something more detailed).

[LambdaDigamma]

I think a resolution that will just smooth the spikes and make the surface look more like a sheet with curves

[billko]

You might try using the graph properties to select a Z-axis range that better suits your needs.  It looks like the graph is auto-selecting the range to give you maximum detail.

[altenbach]

---

@LambdaDigamma wrote:

I think a resolution that will just smooth the spikes and make the surface look more like a sheet with curves

---

You said this is a video. Does the smoothing need to occur in realtime as the data is acquired or is it sufficient to smooth later?

Can you attach a typical 2D sample array? What kind of structure do you expect to remain after eliminating the high frequency noise?

[LambdaDigamma]

OK, this is how it looks after the autoscaling takes some time and stabilizes. It is smoother I know but I just want to get rid of the spices and have like a surface covering the spikes like a tangent maybe.

[LambdaDigamma]

and yes it is a video and the smoothing doesn't have to occur in real time. It can be a post process

[LambdaDigamma]

I was checking the 2d convolution with gaussian kernels, looks interesting. Can you be more specific? Thank you for your time

[altenbach]

You still have not attached any typical dataset.