Two point correlation function for reconstructing image

[Devin-one]

Hi Prof. I want to evaluate the reconstruction effect of SNOW algorithm using two-point correlation function. However, the example in the porespy document only contains the input 3D image, not the reconstructed image. The method I am using is SNOW network extraction (advanced). The last step of the network extraction is

regions = watershed(image=-dt, markers=peaks, mask=dt > 0)
regions = randomize_colors(regions)

I use this regions image to apply the two-point correlation function and only get a horizontal line. I'm guessing this is not a reconstructed image, so how do I get a reconstructed image? Thanks in advance for your reply.

[jgostick]

I am not quite sure what you're hoping to learn by doing a 2point correlation on a watershed image, but I don't think it will work. The image will have values from 0 to N, while the 2point correlation works on boolean images.