how do I calculate area, size and number of colored dots in an image then tabulate them?

numberfiles = length(allFileNames)

circle_count = zeros(numfiles, 1);

% since this is a code for processing multiple images it requires a for

% loop

for k = 1 : numberfiles

fprintf('Processing file #%d of %d : "%s".\n', k, numfiles, allFileNames{k});

% this is the line of code for the images that will be input into a string

% for counting the cells in the image

grayImage = allFileNames{k};

imageData = imread(grayImage);

bw = im2bw(imageData); % Binarize the image

bw = bwareafilt(bw, [10 inf]); % Extract only blobs 10 pixels and larger .

%if you output the data here it would give you the name, size, bytes, class

%and attributes of the image uploaded which is needed for the code to count

%and sum over the number of cells in the image

% Count the number of blobs

[~, numCircles] = bwlabel(bw);

% Save the number of pixels

circle_count(k) = numCircles;

Here is my code that I made for counting the number of colored dots and tabulating them but it is not successfully when applied, maybe the pixels are too small to be counted? Thanks!

Answer 1

[Image Analyst]

I have no idea, not having seen your image. Maybe get rid of call to bwareafilt().

If you have any more questions, then attach your data and code to read it in with the paperclip icon after you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

[jabez]

Here is my image.Thank you, I have read the tutorial and it was very helpful.

@Image Analyst

I removed the bw line and it still did not work.

Goal:

Calculate the area, size and number of dots in 1/2 of the image. i.e. I want to know how does the area, size and number of dots vary on half of the image compared to the other. The code I have written will count the dots and tabulate them but it does it over the whole image not HALF (desired).

Purpose: school assignment (I have already written the code or my attempt of one).

[Image Analyst]

First of all you need to determine how many unique colors there are.

rgbImage = imread('3F5038CD-5F1F-40D2-B990-F0B9A8961AEC.png');
imshow(rgbImage);
[r,g,b] = imsplit(rgbImage);
pixelColors = [r(:), g(:), b(:)];
uniqueColors = unique(pixelColors, 'rows');
fprintf('Total Number of Unique Colors = %d.\n', size(uniqueColors, 1))

Total Number of Unique Colors = 45193.

That's a lot. There are probably some unique colors along the edges due to antialiasing. You might want to visually count how many there are and then use rgb2ind to quantize it, then use ind2rgb to get back to a quantized RGB image with the number of colors you expect. Then you loop over all colors getting a mask for that particular color, then call regionprops to get the area, equivalent circular diameter, etc.

[jabez]

Can I ignore the colors? I am not interested in those features. Just the number, size and area of the dots on the left vs. right. @Image Analyst

[jabez]

@Image Analyst

I included these two lines:

props = regionprops(mask, 'Area');

dotsAreas = [props.Area]

but it gave me the total area(???)

dotsAreas =

630168

I don't want the area of all of the dots, I want the area of the dots in half of the image so I can compare it to the other half of the image. Please advise, thanks!

[Chanille]

So you want to be able count the dots, size and area of half of the image? Right?

[jabez]

@Chanille

Yes

@Image Analyst

Here are the lines that I have added:

%determine area and number of circles

props = regionprops(mask,'Area','Number');

allAreas = [props.Area]

Area will determine the area of each dot in half of the image

Number will determine the number of each dot in half of the image

Then I will put those values into a table

% Stuff arrays into a table variable .

info_table = table(allFileNames(:), circle_count, 'VariableNames', {'FileName', 'CircleCount '});

finalcount = sum(circle_count )

disp(info_table)

But I only know bits of code to make it work but it is not meshing together as a whole and running. Thank you for your consideration in answering my question! @Image Analyst

[Image Analyst]

You can create a mask of all spots by finding the pixels that are not pure white

[r,g,b] = imsplit(rgbImage);
allSpotsMask = (r<255) & (g<255) & (b<255);
[rows, columns] = size(r)
middleColumn = floor(columns/2)
% Get mask on left half
leftMask = allSpotsMask;
leftMask(:, 1 : middleColumn) = false;
% Get mask on right half
rightMask = allSpotsMask;
rightMask(:, middleColumn + 1 : end) = false;
% Measure areas  Overlapping areas just counted once, not twice,
% so a dumbbell has the area of a dumbbell, not of two circles added
% together.
leftProps = regionprops('table', leftMask, 'Area')
rightProps = regionprops('table', rightMask, 'Area')

[jabez]

@Image Analyst

I don't think that answers my question.. can you explain why this possibly would answer my question?

[Image Analyst]

Because you said "Can I ignore the colors?" I made a mask that ignores the colors. Any non-white blob is a foreground pixel in the mask regardless of what the color it.

And because you said "I don't want the area of all of the dots, I want the area of the dots in half of the image" so I made masks of only half the image. Now you can analyze each half by itself instead of analyzing the whole image. You can get the area of each individual cot, or sum them all up to get the area of all dots together.

So I'm meeting both criteria you laid out. Why do you think I don't?