How to add color to the specified blob in a binary image according to the aspect ratio

For example, images with an aspect ratio greater than 2 add red, and images with an aspect ratio less than 2 add green.

I can't find a function about adding color to an area in between. Here is my graph, I want to make the circle red and the bar in the middle green.

Answer 1

[Image Analyst]

Try this (untested):

% Initialization Steps.
clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 14;

grayImage = imread('blobs.png');
[rows, columns, numberOfColorChannels] = size(grayImage);
if numberOfColorChannels == 3
	grayImage = rgb2gray(grayImage);
end
subplot(3, 2, 1);
imshow(grayImage);
title('Gray Image', 'FontSize', fontSize);
axis('on', 'image')

% Binarize the image.
mask = grayImage > 200;
% Delete those blobs that are not completely in the field of view.
mask = imclearborder(mask);
subplot(3, 2, 2);
imshow(mask);
title('Mask Image', 'FontSize', fontSize);
% Label each blob.
[labeledImage, numBlobs] = bwlabel(mask);
props = regionprops(mask, 'MajorAxisLength', 'MinorAxisLength')
aspectRatios = [props.MajorAxisLength] ./ [props.MinorAxisLength]
roundBlobsLabels = aspectRatios <= 2;
roundBlobs = ismember(labeledImage, find(roundBlobsLabels));
subplot(3, 2, 3);
imshow(roundBlobs);
title('Round Blobs', 'FontSize', fontSize);
axis('on', 'image')

wormBlobsLabels = aspectRatios > 2;
wormBlobs = ismember(labeledImage, find(wormBlobsLabels));
wormBlobs = mask & ~roundBlobs;
subplot(3, 2, 4);
imshow(wormBlobs);
title('Worms', 'FontSize', fontSize);
axis('on', 'image')

black = zeros(size(roundBlobs), 'uint8');
rgbImage = cat(3, uint8(255 * wormBlobs), uint8(255 * roundBlobs), black);
subplot(3, 2, 5:6);
imshow(rgbImage);
title('Composite : Worms = red, Round = Green', 'FontSize', fontSize);

[Jiawei Liu]

I tried your method for a day, but failed, can you give a more specific answer? I can't cut red blood cells and bacteria out of shape effectively. and color them all.

[Jiawei Liu]

can you explain the following code? I keep getting an error when I use this line of code

wormBlobs = mask & ~roundBlobs;

[Image Analyst]

Sorry about that. That's what happens if I try to write something off the top of my head without actually testing it in MATLAB. See corrected code above.

That line ANDs the "all blobs" mask with a mask that is "not round blobs", which is white all over except where there is a round blobs where it's black (0). So essentially wormBlobs is all the blobs except the round ones.

[Jiawei Liu]

After attempt your code, I found that if the graphs are mutilated, I cannot sort them and label them. (For example, you masked the mutilated circles near the edges in the image.) My intention is to label the worm figure in the middle with a color (like red), and the other images with similar circles as red. Is there any good way? I don't want to block any mutilated graphics. Please help me, I have been working on this for 3 weeks

[Image Analyst]

Not sure what you're talking about. What graphs? Do you mean images? Just look above at the image. There is nothing mutilated, just round green blobs and red worm blobs. Now you just said you want both of them to be red, or "Like red" and "red". If you want that, it's a simple change but you need to specify whether you want an RGB image or if you want a colormap applied to the image, or if you want red in the overlay above the image, because there are 3 ways to make the blobs red and background black.

[Jiawei Liu]

This is the final result I want, you removed the two circles on the left.This is a binary image, and I want to label the bacteria in the middle green and the red blood cells on the sides red. I have tried using area calculation, but the irregular circle area cannot be calculated. So I asked: Is it possible to use the aspect ratio to distinguish bacteria from red blood cells. You deleted the circle on the left. Not the result I want.

[Image Analyst]

When I calculated the aspect ratio, the slivers on the edges did not have the aspect ratio like you'd expect for a circle, because they're not a circle. They had high aspect ratio as expected. It's extremely common in image analysis to get rid of blobs touching the border because you do not have the entire blob so any measurements you make from a partial blob will not be the same as if you had the complete blob available. If you really want to identify blobs from the boundary as circular or not, it's possible but a lot more complicated, so it's generally not done. It's easier just to take some more sample images and throw out partial blobs.