Counting number of white > 5 Pixel objects in BW image

Hello!

I want to automatically count the particles over 100 microsope images. Due to the sample grind I lost some of my particles and that is why I used the thresholding method according to otsu to make a BW image where my particles and the holes are white and my polymer black. Now im at the "easy part" of counting all the objects. Sadly i just cannot get it running. I tried to delete white pixels for a concentration of less than 5 pixels and simply count with bwlabel the white objects. Is there any other simpler way to count these objects?

please see the attachment.

Thanks!

I found my own solution:

clc
clear all

%Image loading
Oberseite='XXXX';  
oFileName=dir(fullfile(Oberseite,'*.png'));
oNumFiles=numel(oFileName);

for k=1:oNumFiles
    E=fullfile(Oberseite,oFileName(k).name);
    A=imread(E);
    oScans{1,k}=A;
    BW2 = bwmorph(A,'bridge'); %Bridge unconnected pixels
    BW3 = bwareaopen(BW2,5); %Remove objects containing fewer than 5 pixels
    BW4 = imfill(BW3,"holes"); %fills holes 
    [L,Num] = bwlabel(BW4); %Counting white objects in my BW 
    Anzahl_Oben(:,k) = Num %Sum
    
    F=im2bw(A); 
    [r, c]= size(F);
    Size_Oben=F(1:r,1:c);

    %Counting B&W Pixel
    nWhite = sum(Size_Oben(:));
    nBlack = numel(Size_Oben) - nWhite;
    
    %Area related to the particles
    Oben_Flaeche = numel(Size_Oben);

    Oben_Partikelanteil(:,k) = nWhite/numel(Size_Oben);
end

Answer 1

[Image Analyst]

You can use bwareafilt or bwareaopen before calling regionprops.

BW4 = bwareafilt(BW4, 5); % Retain only blobs 5 pixels or bigger in area.
props = regionprops(BW4, 'Area'); % Measure areas.
allAreas = [props.Area] % Extract from structure array into a vector.
numParticles = numel(props) % Count the number of blobs in this image.

[CJ]

Thank for your answer. Isnt bwareafilt for retaining only the 5 objects with the largest areas? Thats why I used bwareaopen

[millercommamatt]

You're correct about bwareafilt.

This syntax will work for filtering by pixel area:

BW4 = bwareafilt(BW4, [5 9e9]); % 9e9 or just some big number to set the max area larger than your largest expected value.

[Image Analyst]

Yeah, sorry about that. @millercommamatt thanks for the fix.

What I had meant to use was bwareaopen instead of bwareafilt:

BW4 = bwareaopen(BW4, 5); % Retain only blobs 5 pixels or bigger in area.

So either function should work as long as you call them before regionprops.

It looks like you just wanted the total area of all blobs summed together. regionprops will give you the area of each blob individually. If that's the case you can do either

totalNumWhitePixels = sum(allAreas);

or

totalNumWhitePixels = nnz(BW4);

Let me know if this works for you.

[CJ]

Thank you again. You are right, I want to count the white objects and get the total area. Its seems that my skript does already that. So I will let it untouched

Answer 2

[millercommamatt]

You should be able to to something like this. I'm waiting on a slow scatter interpolation on my client to finish so I haven't tested this so the syntax might night be quite right, but the idea is.

stats = regionprops(BW); %BW is the array of your binary image
num_object_get5 = sum(stats.Area >= 5);

[CJ]

Thank you for your answer! Sadly i get an error that i use too many input arguments for this