I managed to label and get set of pixels for every single object of interest in the image.
Now I'm looking into clustering and am not sure what features are worth using. I currently have two in mind
$x_1 $= Areas of objects
$x_2 $= Mean Grayscale Pixel Intensity
Now i'm not sure if I should get more features in order for clustering to work better. I was thinking maybe feature 3 be a ratio of feature 1 and 2, or get the histogram of each object, and find a parameter to describe the histogram. I am able to take any of these features of reigionprops as well, but I'm not sure how much useful would half of them be.