python - Image processing approach -

in following greyscale image,i trying identify objects have manually labelled in red. have suggestions of how this?

i have attempted use gaussian blur , thresholding cannot exclusively identify these particles. edge detection approach? suggestions welcome.

your images suitable target machine learning.

• i created following probability map using trainable weka segmentation plugin imagej/fiji (taking second image input):

you should able objects simple thresholding , possibly subsequent size filtering.

• you might want try ilastik, free software combines pixel classification , object tracking power of machine learning.

edit:

this how proceeded using trainable weka segmentation:

1. in settings window, activated more features, set sigma range 4 32, , named classes "objects" , "background":

2. i created freehand line traces, added them respective classes, , clicked on "train classifier". (the creation of feature stack takes while first time run training, refining classification takes less time because classification need run.)

3. to probability map, click on "get probability".