Dear Vladislav,

Here my answer in brief:

I would recomment to dublicate the picture and transform it into a 8 bit image as the first step. Then you should run a edge detection. After that you can threshold the dark areas (pores) that does not have to be perfect, since you will use the fill wholes option in the analyze particles modul. I would recomment to use the following settings: exclude 20 pxl area objects, add to manager, include holes.

If that is already helping you to head you in the right direction great - otherwise I would also be able to show my workflow approach in a pdf - if needed.

kind regards

Soenke

A possible method would be to mesure the local thickness of all boundaries and pores together. Next you apply a thikness threshold, in effect saying that structures below a certain thickness are not pores.

For example, using the local thickness plugin by Bob Dougherty with reverse threshold of 160 and extracting thickness regions larger than 4.5 pixels results in a pore area fraction of 0.0083 (0.83%). This is illustrated in the attached image.

Local thickness plugin: https://www.optinav.info/Local_Thickness.htm

Reference: "A new method for the model-independent assessment of thickness in three-dimensional images" T. Hildebrand and P. Rüesgsegger, J. of Microscopy, 185 (1996) 67-75

Likely the best way to do this, based upon experience, would be to use Computer-based X- ray Tomography ( or CT). One can easily analyzed porosity, pore size distributions, comparison between breads, consistency, poor size differences thruogh- out the loaf, as well as cell wall thicknesses and their distribution through out the loaf, and comparisons between differing bread formulations and baking conditions. One can analyze crusts (top, bottom, sides and ends), or the impact of various additives. All of this can be done in three- dimensions, in bulk or with virtual sections

Too, with the right instrumentation, this can be done with micro or even nano resolution. Something worth consideration.

Sönke Weinert Thank you for your reply. I turned on the edge detection, but I'm not quite sure what to do next. Could you please describe the procedure in more detail?

Dear Vladislav S. Shikalov ,

Please find attached a example workflow. Due to the answer of Var. St. Joer I'm not really sure if that is what you are looking for to measure.

If not please let me know.

Also: If anybody finds a typo in the file - you are well come to keep it - I have more than enough of them.

Kind regards Soenke

Usually, I would also recomment to click the exclude edges option in the particle analyser.

I would recommend using binary mode. No gray levels. Then measure the total area taken by both imaged phases and subtract one from the other. You will need to have several analyzed images. Then average all data from each phase. The more images that are analyzed, the more accurate your averaged data will be. This mythology is tedious and a little time consuming., but it is a very good, reproducible, and will give you your answer.