0

I have ring-like binary images with varying thicknesses. I want to calculate its thickness by calculating the distance between the inner and outer boundaries.

enter image description here

In many places, people used distance transform to do this. Ideally, if I take the distance values in the medial axis, it will give me the thickness by distance*2.

However, in my case, the shape is very uneven; hence the medial axis can branch. But since the 3D structure I am analyzing is tube-like, I need a thickness value that I can project to the outer boundary to visualize. This means the thickness values should be strictly on a ring-like medial axis.

I am wondering if we can do the distance transform from one boundary to another instead of starting from boundaries and going toward the center.

Savindi
  • 123
  • 1
  • 8
  • So... the required task may be... for every point on the outer boundary, finding the corresponding point on the inner boundary? – fana Apr 20 '23 at 01:44
  • I don't know if it works well and/or how to implement it, but I imagined the following story. Points on the outer contour are connected with neighboring points by springs. Considering the distance transform result as height field, all points start climbing at the same time. They will reach the "center of the thickness". Then, descend inward to find the corresponding point for each. "thickness" can be the length of the path followed by each point, or may be able to be approximated simply by the distance between correspondences. – fana Apr 20 '23 at 01:58
  • Thanks @fana. Yes, it is finding an inner boundary point for every point in the outer boundary. When we go ahead with what you have mentioned, it still gives that unwanted branching effect because we ultimately take the point in the center line to calculate the path of the point... – Savindi Apr 21 '23 at 00:12

0 Answers0