Droplet size calculation of a spray image

Hi predecessors,

I have this kind of spray image and I want to calculate the droplet size and distribution of the particles, and wondering if there is a way to do that on MATLAB so that I can learn from it. Thanks!

Here are what I do now, I really appreciate you can help to point out the errors or mistakes,and if you are me, what code will you use to realize this function.

clc
clear all;
image=imread('E:\picture_test\heitu.tiff');  
jqimage=image(2000:2500,2000:2500);  
imshow(jqimage,[0 4000]);title('jiequ');

[gradthresh,numofiter]=imdiffuseest(jqimage);
diffusedImage=double(imdiffusefilt(jqimage,'GradientThresh',gradthresh,'NumberOfIterations',numofiter));
imshow(diffusedImage,[0 4000]);

[testx,testy]=size(diffusedImage);
array_x=zeros(501,501);  
blank_x=zeros(501,501);
y=zeros(1,501);   

for i=1:testx     
    testx_temp=diffusedImage(i,:); 
       for i1=2:testy-1      
           y(i1)=abs((testx_temp(i1+1)-testx_temp(i1-1))/2);   
       end    
    array_x(i,:)=y;  
    [pks,locs]=findpeaks(array_x(i,:),"MinPeakHeight",40);
    blank_x_i=blank_x(i,:);  
    for i2=1:length(locs);
        blank_x_i(locs(i2))=1; 
    end
    blank_x(i,:)= blank_x_i; 
end

array_y=zeros(501,501);   
blank_y=zeros(501,501);
x=zeros(501,1);   
for i=1:testy     
    testx_temp=diffusedImage(:,i);  
       for i1=2:testx-1      
           x(i1)=abs((testx_temp(i1+1)-testx_temp(i1-1))/2);   
       end    
    array_y(:,i)=x;  
 
    [pks,locs]=findpeaks(array_y(:,i),"MinPeakHeight",40);
    blank_y_i=blank_y(:,i);   
    for i2=1:length(locs);
        blank_y_i(locs(i2))=1;   
    end
    blank_y(:,i)= blank_y_i;  
end
figure;
imagesc(blank_x);
hold on;
imagesc(blank_y);
add_ab=blank_x+blank_y;
imagesc(add_ab);

figure;
binaryimage=im2bw(add_ab);
imshow(binaryimage);
binaryimage=imfill(binaryimage,'holes');
imshow(binaryimage);title('binaryimage');
figure
labeledimage=bwlabel(binaryimage,8);
imshow(labeledimage);
coloredlabels=label2rgb(labeledimage,'hsv','k','shuffle');
hold on;
imshow(coloredlabels);
blobmeasurements=regionprops(labeledimage,add_ab,'all');
numberofblobs=size(blobmeasurements,1); 
boundaries=bwboundaries(binaryimage);   

numberofboundaries=size(boundaries,1);

for k1=1:numberofboundaries
    thisboundary=boundaries{k1};
    plot(thisboundary(:,2),thisboundary(:,1),'g','LineWidth',2);
end 
for k2=1:numberofblobs
    blobarea=blobmeasurements(k2).Area;
    blobecd(k2)=sqrt(4*blobarea/pi);
      fprintf('%d\n',blobecd(k2))
end

[Image Analyst]

This

jqimage=image(2000:2500,2000:2500);

does not work with the particle.jpg image you attached. It's not big enough.

[Jason]

Thanks professor, I know this problem, now it is only for testing.Could you please help write a program that can implement this function?it may be better to learn.Thanks a lot

Answer 1

[prasanth s]

Use 'kmeans' for segmentation and 'regionprops' to get properties of particles,

image=imread('particle.jpg');  
jqimage=rgb2gray(image);  
[idx,C]=kmeans(jqimage(:),5);

Warning: Failed to converge in 100 iterations.

c_image=reshape(idx,size(jqimage));
droplets=c_image==1;
imshow(droplets);

stats=regionprops(droplets);
average_area=mean([stats.Area])

average_area = 3.1781

[edit: attached image and ran the code]

[Jason]

Thanks for your kind help,I'll try it

[Khai]

@Jason Could you please help me to provide full code. Thank you so much for your supprot

[Image Analyst]

kmeans is not the best function to use here. It is basically going to threshold the image into 5 different brightness ranges. In fact class/range 1, which he uses to form the segmented image, is unknown - class 1 may refer to different intensity ranges and hence different particles every time (will not necessarily always be the darkest particles).

Answer 2

[Image Analyst]

Try imbothat bottom hat filter.