Finding Circles and Their Areas

Using some code I have written, I can take an image, remove the background noise, and semi-count up the total number of circles in the image. The problem I am running into is as follows. Bellow is a reference image.

As you can see the code correctly count the majority of the droplets, using the code

[centers, radii] = imfindcircles(bw,[20 20],'Sensitivity',0.99,'Edge',0.03);

I don't want to increase my sensitivity any further, because then it starts counting circlets that aren't droplets.

I also tried counting the number of objects in the picture, but the problem then becomes that it starts counting that thin sliver of white space on the very right edge of the image. Also this does not help me find the area of all of the circles. Right now I am using the code

cc = bwconncomp(bw, 8);
dropdata = regionprops(cc, 'Area');
dropArea=[dropdata.Area];

But since this is only counting the are of the objects, it counts the area of the entire stream and the drop of the first object, while I just want it to count the area of the drop on the end. Any help would be greatly appreciated.

Answer 1

[Image Analyst]

Try watershed to split it apart.

[Nathan]

When I try that it just gives me a solid blue image.

L = watershed(bw);
rgb=label2rgb(L,'jet',[.5 .5 .5]);
figure()
imshow(rgb)

[Image Analyst]

Post a couple of original, non-annotated, images and I'll look at them over the weekend if I get time.

[Nathan]

After tweaking around with the watershed example, I think I actually figured it out. I can get a nice segmented image as seen below:

The only problem with this being, it segments the stream into two separate pieces. Any idea what could be causing this? And on another image, it doesn't want to segment one of the drops from the white line on the right of the screen.

The code I used is as follows:

D = -bwdist(~bw);
Ld = watershed(D);
bw2 = bw;
bw2(Ld == 0) = 0;
mask = imextendedmin(D,2);
D2 = imimposemin(D,mask);
Ld2 = watershed(D2);
bw3 = bw;
bw3(Ld2 == 0) = 0;

[Nathan]

For removing the white line on the left I just decided to drop the last two columns of my bw image, so now that problem is solved.

[Image Analyst]

How would cropping off the last two columns on the right get rid of the line on the left?