1
$\begingroup$

I am looking for open source implementation of thinning algorithm in 3D.

Thinning algorithm: Convert binary shapes to 1-pixel wide lines, without shortening it or breaking it apart. For example, the thresholded version of hand written or printed alphanumerics can be thinned for better represetation and further processing.

This particular thinning algorithm I am looking for does not disconnect an object or split an object up into multiple parts. Moreover, it does not shorten the shape.

Are there such source codes available? If 3D is not available, 2D is fine too. I can take it from there.

UPDATE:

Somebody has requested me to elaborate on what am I after. Here: http://www.mip.sdu.dk/ipl98/how_to_use_ipl98/tutorial/node52.html There are several images in the link here on thinning and skeletonization. I can see that thinned algorithm and skeletonization algorithm produce similar yet different result. But what I am after is to have all shapes in a binary image to become thinned without shortening the shape. To clarify things further, here is what I wanted:

This is original image:

original image

This is skeleton image:

skel

This is thinned image:

thin

As you can see, skeleton image and thinned image have different results. Thinned image has lesser lines and produce an outline that looks more like the thinned original. But these images are in 2D, I am looking for an algorithm to do thinned binary image in 3D.

$\endgroup$
2
$\begingroup$

If you would be happy with 3D skeleton and are looking for a Java implementation, there a popular image processing package in Life Science called ImageJ. In a distribution of ImageJ called Fiji, you will find a solid implementation by Ignacio Arganda-Carreras.

Check here: http://fiji.sc/wiki/index.php/Skeletonize3D. You will find a link to the source as well as links to tutorials.

Here is video describing how to use it from with the Fiji software: http://fiji.sc/wiki/index.php/Skeleton_analysis_video_tutorial

$\endgroup$
0
$\begingroup$

The Insight Toolkit (ITK) has an implementation of the euclidean signed distance transform. It is fairly well know that the 3d skeleton is a level curve of the distance transform of a binary object.

$\endgroup$
  • $\begingroup$ But distance transform gives us skeleton, not thinning, does it? $\endgroup$ – Karl Jan 17 '12 at 4:40
  • $\begingroup$ Can you elaborate on the difference between the skeleton and thinning ? $\endgroup$ – nav Jan 17 '12 at 5:30
  • $\begingroup$ I've edited the question to elaborate my question further. :) $\endgroup$ – Karl Jan 18 '12 at 5:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.