Is the return value of filters.local_thickness a diameter or a radius?

[LuvMangorange]

According to the documentation, the returned value is the radius, with a minimum value of 1, which corresponds to the image resolution. This implies that the minimum pore diameter is twice the image resolution. In reality, the minimum pore diameter should be constrained by the image resolution, meaning the minimum pore diameter should be 1.

[jgostick]

It is radius. The distances in digital images are measured from the center of a pixel to the center of the nearest solid/background pixel. So if your image has 'pores' which are just a single pixel wide, then it has a radius of one pixel. I understand your point about such a pixel having a diameter of 1, but this is a limitation of digital images. Another way to look at it is that the true void/solid boundary is not perfectly defined, so the only thing we can say for certain is that the center of neighboring pixel is solid and the boundary occurs "somewhere" in the middle. This limitation of course becomes increasingly irrelevant as the pores become larger.

[LuvMangorange]

Thank you for your kind response. What you said is likely correct, as the results I obtained using another library (ndimage.distance_transform_edt(self.porous)) are also largely consistent with those from filters.local_thickness.