Count using area sizes instead of COLOR CONTRAST

nomipakistanli

Dear members, and specially Yuri,
I would like to thank on behalf of all members you have helped till now. 
I have this image, attached, in which there are two main components but unfortunately have same tones. I can not use the color distinction to acquire counting or other data. is there any way i can count the size or at least the percentage of the smaller circles (new phases) in the micrograph. 
NOTE: Small black dents in the smaller circles can be ignored..

Regards and thanks.
NUMAN

mbatchelor

2021-02-09-184252

NUMAN --

I have taken a different approach to your challenge.

Please review the IMAGE(S) below and the PDF FILE attached as

    2021-02-09-190051.pdf

and contact me if this looks like a promising approach.

Thanks.

-- Matt

-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*

YuriG

Hi NUMAN,

The image contains only the boundaries of the objects to measure, this is a tricky task. The approach would be to segment the boundaries using smart segmentation and use "Fill Holes" option to fill the closed objects. I've attached the IQO file that attempts to do that, the result looks like this (the percent area is reported in the stats pane) :

The problems with this approach is that it will work only on sparse distribution of objects with well-defined boundaries. If boundaries are opened - the object will not be counted. If multiple objects grouped in clusters, the background area inside the cluster will also be counted as object. So, it's difficult to pick the right segmentation, which would close weak boundaries and at the same moment prevent counting background areas inside connected object clusters.
You may try to tweak my recipe to get better results, but having 2 phases to be different shades or colors would be a better approach (you may try using different illumination, polarization,... to find differences in phase colors).

Yuri

nomipakistanli

Dear Yuri Thank you very very much for the efforts. i have just been able to collect the data for all components. will try to use the file u sent.