How to find the dark pixels in an image?

Hello. I have a lot of images like this one:

From the whole shrimp, I would like to determine the area of dark pixels as a percentage of the visible surface. Does anybody have an idea how to do that? My idea is to first define/find the desired color and then filter the image for that defined color.

Thanks Falamanki

Answer 1

[Bshara Murr]

I = imread('shrimp.bmp');

N = I(:,:,3)>100;

I(:,:,2) = I(:,:,2)>100;

Where the second line treats each color channel separately, 3 represents the blue color, you can use 1 and 2 for red and green the 100 is the threshold that you can change as you with, you can use < as well, this gives you a bw image of the desired channel. the third line returns an rgb image but with a certain color channel thresholded but it will alter your white background as well.

[Bshara Murr]

Just a question out of the topic. Are you lebanese?

[Al Falamanki]

No, I am not Lebanese, but I know the restaurant "Falamanki" in Beirut... ;-)

[Bshara Murr]

Aha okay. Did you find what you were looking for?

[Al Falamanki]

I am working on it, but so far, i think will get a satisfactory solution. Thanks !!

Answer 2

[Walter Roberson]

You could try rgb2gray, then imbinarize using some brightness level you consider to be "dark", and count the number of 0 in the result.