How to count the amount of small squares in this picture?

[xie]

I mean I want to count units quantity as below:

Answer 1

[DGM]

The given answers have problems with accuracy. One way to help with discrimination is to do some sort of template matching.

% the image
inpict = imread('https://www.mathworks.com/matlabcentral/answers/uploaded_files/1821039/image.png');
inpict = im2gray(inpict);

% the sub-image template
template = imcrop(inpict,[22.51 22.51 14.98 13.98]);

% get a correlation map describing 
% where copies of the template are
C = normxcorr2(template,inpict);
mask = C > 0.5;
mask = bwareaopen(mask,9); % get rid of any specks
imshow(mask)

% count the peaks
N0 = bwconncomp(mask).NumObjects

N0 = 2925

Is that number plausible? The image isn't that large. We can just manually verify that there are 45x65 = 2925 copies of the template object. We can also do that programmatically:

% number of rows of objects
nv = smooth(max(C,[],2)) > 0.5;
nv = bwconncomp(nv).NumObjects

nv = 45

% number of columns of objects
nh = smooth(max(C,[],1)) > 0.5;
nh = bwconncomp(nh).NumObjects

nh = 65

% total
N = nv*nh

N = 2925

As is, the other examples produce an overestimate due to the occasional blob between objects, or content along the image boundary. When I tested them, the estimates were higher by about 100 or so.

[xie]

Great!

This way get rid of the noise. I can get true result!

Answer 2

[KALYAN ACHARJYA]

Approximate Way:

image_input=imread('image file name');
image_bw=imbinarize(rgb2gray(image_input));
cc=bwconncomp(image_bw);
number=cc.NumObjects;

However, you can obtain the exact value by incorporating additional morphological and logical statements.

[xie]

Thank you KALYAN!

Answer 3

[Image Analyst]

See my Image Segmentation Tutorial in my File Exchange:

Image Analyst's File Exchange

It's a generic, general purpose demo of how to threshold an image to find blobs, and then measure things about the blobs, and extract certain blobs based on their areas or diameters.

Answer 4

[Catalytic]

BW=imbinarize(im2gray(imread('image.png')));

BW([1,end],:)=0;  
BW(:,[1,end])=0;
BW=imopen(BW,ones(2)); 

nv=numAlong(BW,1)

nv =

45

nh=numAlong(BW,2)

nh =

65

total=nv*nh

total =

2925

function out=numAlong(BW,dim)

  s=reshape( sum(BW,3-dim),[],1);
  
  m=bwareaopen(s>max(s)/3,8);

  out=bwconncomp(m).NumObjects;

end