Extrema of a PointCloud surface

Hi,

I have a set of data points in 3D space (a matrix of (n x 3), in which each row corresponds to a point) that are extracted from a stl file and all together describe an irregular surface. I want to find the extrema of this surface. Is there anyway to do this in matlab? Attached is a sample file.

I know using the approach described in the following post we can find the extrema but the problem is that my data points are not continous and I cannot use meshgrid.

https://www.mathworks.com/matlabcentral/answers/277512-how-to-find-peaks-in-3d-mesh

Thanks.

[Nick Keepfer]

I think you are asking for what is known as a convex hull.

If so, there is a built-in function inside Matlab called convhull, which would be ideal for you.

https://www.mathworks.com/help/matlab/ref/convhull.html

[Memo Remo]

Hi Nick,

Thanks for the reply.

No. My geometry is a surface not a volume.

[Nick Keepfer]

For the extrema in the z axis you can use

scatter(x,y,20,c,'filled')

Here, x, y are the coordinates of your points in x and y, and the colour c can be the height in z.

[Memo Remo]

Thanks for your help. It seemed very interesting.

However, it converts the 3D data to 2D and finds minimum or maximum z values, which doesn't correspond the local extrema.

[darova]

Can you calculate normal vector of each face? Find vector v = [0 0 1]

[Memo Remo]

Thanks Darova,

I think I should be able o do that. However, it is not very easy to do so.

You mean I should search for those locations that has minimum variations in the direction of the normal vector?

Answer 1

[darova]

Here is the idea:

• use griddata to build a surface
• use imregionlmax to calculate extrema

[Memo Remo]

Great!

"griddata" was the function I was looking for.

Thanks!