How to fit the cylinder problem with spatially dispersed data?

1.Here I have 64 coordinates of the irregular cylinder contour, that is, I divide the contour into 8 sections, a total of 64 grids, and I want to fit the cylinder according to these 64 spatial coordinates. At present, only the center of gravity of each section is used for fitting, which is not effective.

2.First, we want to carry out cubic spline interpolation for the scattered points, interpolation four sections between each two sections, fitting the section circle through the least square method, and fitting the cylinder.

3.The center of each section is expected to be connected with the axis

4.For example, the least square method, the least cylindrical inclusion method

Try to fit the pattern in the drawing