How to isolate particles from similar background?

[shalopez]

I am grad student in geology getting into Image J for my thesis project and would like to accomplish a few things. I am analyzing a rock unit that decreases with depth laterally and has changes in compaction with burial depth. The rock is made primarily out of pellets with an ovalish to ellipsoid shapes.

I would like to quantify the pellet shapes and sizes using the circularity, roundness, and area tool in the 'measure' feature in Image J. I would also like to record the color of the pellets to see if this plays a factor in compaction (and ultimately different mineralogy of the pellets). However, Image J only measure intensity.. Ive seen a few plug ins available such as RGB measure, but i only want one row for all measurements, not 5 like this one gives. Is there a way to include this average of each RGB for a selected ROI into the one line of measurements for area, circularity, ect...It seems that if i just used image j's standard pixel measure of intensity, it wouldnt be a good proxy for visual color of the pellets.

Another problem is that the images taken from thin sections have a lot of areas between the pellets and the matrix that look the same and are made up of similar minerals. Since i am not an image analysis expert I find it would be hard to separate out the pellets since some times this can be an interpretive task. Attaching an example of what i am working with. I imagine i will have to select each pellet individually since some are hard to pick out and some pellets are merged.

Can anyone help me out with a process?




Thanks!