3D Fit of Centreline

I've got a 3D data set (shown below) and I want to fit it with a line, the points of this line fit need to be equidistant apart.

Any tools able to do this? Or methods of how to achieve this?

Many thanks, Will

Answer 1

[Image Analyst]

Have you tried a Savitzky-Golay filter on the x, y, and z independently? Attached is a demo of sgolayfilt() in the Signal Processing Toolbox.

Answer 2

[Image Analyst]

Do you want to smooth (regress) those points to get a smoother path? Or interpolate additional points in between those points while keeping those points where they are?

Of course we should not forget the venerable John D'Errico's interparc : http://www.mathworks.com/matlabcentral/fileexchange/34874-interparc. It sounds like just what you want to do if you want to keep those training points fixed and interpolate evenly spaced points along the curve.

[William]

After this step I will be doing analysis of the curve which requires equally space points along the line. So I need a mixture of the two, smoothing with additional points.

[William]

I'll try a S-G Filter followed by interparc

[William]

Looking tasty. Could we calculate a standard deviation or some other error measure against the original points?

[William]

Simple enough - Minus filtered from original data and take statistical profile of this.

Thanks very much for your help :)