Image processing: Find curvature in a photomicrograph

Dear Sir or Madam,

Dear colleagues,

I am investigating the curvature of fibre bundles in a specimen produced out of carbon fibre reinforced plastic. The curved fibre bundle itself lies in a curved specimen (a segment of a tube cut under 45° to its rotational axis). I attached a typical photomicrograph and an edited picture that shows what I am interested in. The aim of my work is to receive either the (more or less) exact curvature of the fibre bundle. Or at least a possibility to “insert” points whose (pixel-exact) coordinates I can export and later on set up a polynominal function through these points.

Do you have any suggestions on how to get the curvature of the fibre bundles? Is Matlab therefore suitable?

I am looking forward to hearing from you and thank you for all constructive suggestions.

Best regards,

Felix

[Image Analyst]

It's been 5 years. Do you still need an answer? If so do you just want a radius of curvature for the whole white bundle (easy) or of each fiber in the bundle (a little tricky but I do have a demo)?

[felix.b]

You are right, it's been quite some time. I do not need an answer anymore because I found a solution. However, I am (always) interested in learning new things. That is why I appreciate if you can share your solution. Time is not important anymore. So, if you have to solve other tasks before, please do so, and then come back as soon as you are free again.

[Image Analyst]

What was your solution? If you don't have thousands of these to do, I'd probably just use drawpolygon or drawpolyline to manually trace them. You'd probably spend less time drawing them manually than developing an automatic way, which might not even be as accurate since it's not clear where the fibers are, like where they merge or separate or start or end.

[felix.b]

Good to read about your idea and the Matlab routines that you mentioned.

My solution was a similar to your idea. I set a scale (in Millimeters) and then pick locations along the fibres' paths in the open source software Fiji/ImageJ (https://imagej.net/software/fiji/). I extracted these points and processed them quite automatically in Matlab using my own and speciliazed script then. Please see the following reference https://athene-forschung.unibw.de/151585 (especially page 112ff) for further information. Although it is in German, I am sure you will understand my approach. If you further interest please do not hesitate to contact me.