jacopoantonello/mshwfs

Modal Shack-Hartmann wavefront sensor toolbox for MATLAB

mshwfs's Issues

Centroid Function: Inconsistency in Variable Convention and Output Format

In the centroid function, there is a swap of the X and Y variables in both the input and output. Although the fragment "Centroid works on column major order" is present, it does not explicitly state that this function returns positions in the format [Y, X]. Especially considering that all code in centroid.m suggests it follows the convention [X, Y]. This change in convention also complicates swapping the centroid method with others, as other methods typically use the convention [X, Y], thus reducing the code's flexibility. To rectify this and provide clarity:

[w, h] = size(im); -> [h, w] = size(im);
Size returns length of dimensions in order Y, X, Z...

[yy, xx] = meshgrid(1:h, 1:w); -> [xx, yy] = meshgrid(1:w, 1:h);
Meshgrid input and output is [X,Y] = meshgrid(x,y)(https://se.mathworks.com/help/matlab/ref/meshgrid.html#d126e1020611)

In the function outside of centroid in shwfs_make_fine_grid.m, there is a change:

dd = shstruct.centroid(subimage, level);
centres(i, :) = [cc(1)+dd(2)-1, cc(3)+dd(1)-1];
It is swap of x and y from dd. dd(1) = X and dd(2) = Y, so it is counterintuitive to add cc(1)+dd(2) (X+Y) and cc(2)+dd(1) (Y+X).
After changes in centroid.m it should be:
centres(i, :) = [cc(1)+dd(1)-1, cc(3)+dd(2)-1];

The relationship between the reconstructed wavefront and the actual wavefront.

Hi author, I would like to ask, the range of the wavefront reconstructed from this project is only the size of the unit circle, but the size of the whole pupil is not necessarily the unit circle. So I would like to ask how to map the reconstructed wavefront to the actual measured wavefront?

Error using rectangle in calibration.

I am using .bmp image for calibration after converting it into .mat. When I reduce the shstruct.coarse_grid_radius, i.e.radius for coarse grid from 16 to 10, for instance, the error that comes is, "any non finite? 1". It happens with some images. Most of the images can manage coarse grid radius of a wide range like (10,16). Some can take only above 15. I guess, one of the centers in shwfs_make_fine_grid code takes non-finite value, which one, I don't know. What to do to handle this error?