HELP!! No matter how big I set the threshold THE OBJECTS I DON'T WANT ARE NOT HIGHLIGHTED!!!

% Read image as RGB
I = imread('TEST.TIF'); %including the pull path is better
% Convert to binary
BW = imbinarize(rgb2gray(I));
% Get rid of white border
BWConv = bwconvhull(~BW);
BW = BW & BWConv;
% get region centers and major axis, and orientation
stats = regionprops('table',BW,'BoundingBox','MajorAxisLength','MinorAxisLength');
% Compute length/width ratio
stats.LenWdRatio = stats.MajorAxisLength ./ stats.MinorAxisLength;
% Set threshold and determine which objects
% are greater than the threshold. 
thresh = 10000.0; 
stats.isGreater = stats.LenWdRatio > thresh; 
% Show bounding box for objects that met the threshold
figure()
h = imshow(BW); 
axis on
hold on
% Plot red rectangles around accepted objects
arrayfun(@(i)rectangle('Position',stats.BoundingBox(i,:),'EdgeColor','r'), find(stats.isGreater)); 
% Plot yellow rectangles around rejected objects
arrayfun(@(i)rectangle('Position',stats.BoundingBox(i,:),'EdgeColor','y'), find(~stats.isGreater)); 

I dont want those big black circles in my image but I can't determine their threshold to highlight them. PLEASE HELP!! This is really doing me in a good one in a bad way, thanks!!

It should look like this (yellow is rejected) where I've circled some of the rejected objects:

[Rik]

You know a lot more with that color image than just the area of the objects.

Why are you not using the fact that you know that you want a (mostly) black region with a yellow border? Why don't you first try to find black areas, after which you can either check the size or confirm they have a yellow border?

[Laura]

@Rik

That sounds like a great idea actually, but how do i do that?

[Steven Lord]

What results do you receive when you run your code? And how do you distinguish between similarly shaped and sized objects in your image where some are rejected and some aren't? I'm particularly looking at the rejected circle in the lower-left corner of your image. There are four other circles (or circle-like areas) around it (using clock position notation at 12 o'clock, 2 o'clock, 4 o'clock, and 8 o'clock) that look awfully similar to that rejected circle. Do you expect those to be rejected? Why or why not?

[Laura]

@Rik This code I wrote to do that is not succesful:

% Read image as RGB
I = imread('testimage.png');
% Convert to grayscale
Igray = rgb2gray(I);
% Threshold the image to get binary mask of black regions
BW = imbinarize(Igray, 'adaptive', 'ForegroundPolarity', 'dark', 'Sensitivity', 0.2);
% Get rid of small objects and fill holes
BW = bwareaopen(BW, 500);
BW = imfill(BW, 'holes');
% Detect yellow borders
se = strel('disk', 10);
yellow = imopen(Igray, se);
yellowMask = imbinarize(yellow, 'adaptive', 'ForegroundPolarity', 'dark', 'Sensitivity', 0.2);
% Remove small yellow regions
yellowMask = bwareaopen(yellowMask, 200);
% Get the black regions with yellow borders
blackWithYellow = BW & yellowMask;
% Get region properties of black regions with yellow borders
stats = regionprops('table', blackWithYellow, 'BoundingBox', 'Area');
% Check if the black regions have a minimum area of 5000 pixels and plot bounding boxes
figure;
imshow(I);
hold on;
for i = 1:size(stats, 1)
    if stats.Area(i) >= 5000
        bbox = stats.BoundingBox(i, :);
        rectangle('Position', bbox, 'EdgeColor', 'r', 'LineWidth', 2);
    end
end

[Adam Danz]

@Laura see my response to your comment in this thread which appears to be where you got the code used in your question.

[Jan]

@Laura: It is useful in general to avoid vague descriptions as "is not successful", but explain, what you observe and what the problem is. This information is valuable and you have it already. Sharing it with the readers makes it easier to suggest a solution.

[Laura]

@Steven Lord

When i run this code:

% Read image as RGB
I = imread('colortestimage.png'); %including the pull path is better
% Convert to binary
BW = imbinarize(rgb2gray(I));
% Get rid of white border
BWConv = bwconvhull(~BW);
BW = BW & BWConv;
% get region centers and major axis, and orientation
stats = regionprops('table',BW,'BoundingBox','MajorAxisLength','MinorAxisLength');
% Compute length/width ratio
stats.LenWdRatio = stats.MajorAxisLength ./ stats.MinorAxisLength;
% Set threshold and determine which objects
% are greater than the threshold. 
thresh = 2.0; 
stats.isGreater = stats.LenWdRatio > thresh; 
% Show bounding box for objects that met the threshold
figure()
h = imshow(BW); 
axis on
hold on
% Plot red rectangles around accepted objects
arrayfun(@(i)rectangle('Position',stats.BoundingBox(i,:),'EdgeColor','r'), find(stats.isGreater)); 
% Plot yellow rectangles around rejected objects
arrayfun(@(i)rectangle('Position',stats.BoundingBox(i,:),'EdgeColor','y'), find(~stats.isGreater)); 

i get this:

[Laura]

@Steven Lord the code does not let me select these areas (i circled some of them) that i want. The goal is to substract out those yellow like objects I circled and select for everything else:

[Rik]

How were you planning to subtract those regions? They are not currently selected, so how did you plan on removing them again?

I think you need to go back to the drawing board. Start with describing steps (in plain English) describing what you want each step to do. Those are your comments. Then try to implement this by translating the English comment to Matlab.

[Laura]

@Jan You are absolutley right. Thank you for your correction, I will avoid vagueness in the future.

@Rik That is my question. how can I substract those regions out? How can I select them?

1. I have this image with black objects in the foreground. I want to select those black objects and substract them from the image. How do I do this?

Well, I made a code that would select the black objects and substract them to the color image example:

im = imread('colortestimage.png');
im_gray = rgb2gray(im);
bw = imbinarize(im_gray, 'adaptive', 'ForegroundPolarity', 'dark', 'Sensitivity', 0.5);
bw_filled = imfill(bw, 'holes');
bw_cleared = imclearborder(bw_filled);
% Keep only the large and connected components
bw_large = bwareafilt(bw_cleared, [500, Inf]);
% Invert the binary image
bw_inverted = ~bw_large;
% Subtract the black objects from the original image
im_subtracted = im;
im_subtracted(repmat(bw_inverted,[1,1,3])) = 0;
% Visualize the results
figure;
subplot(1,2,1);
imshow(im);
title('Original Image');
subplot(1,2,2);
imshow(im_subtracted);
title('Subtracted Image');

but the result was a fully black image. When I expected just the large black circles and tube shape black areas to be selected and substracted. Can you suggest how I can do what I expected? Thanks.

[Rik]

What would those areas be replaced with? You really need to describe in words what you want. What you keep posting is an end goal and not the steps. The fact that it is unclear what you want doesn't help.

[Laura]

@Rik I want to measure the white blobs and the space in between them in an image. However, there are black areas in the image that interfere with my measurements. To eliminate this interference, I need to subtract these black areas from the image. Therefore, I need to specify which areas in the image are black and remove them, so that I'm left with only the relevant information to measure.

% Read image
im = imread('image.jpg');

% Convert to grayscale if necessary
if size(im,3) > 1
    im = rgb2gray(im);
end

% Threshold image to create binary mask
mask = imbinarize(im, 'adaptive');

% Remove small objects from binary mask
min_size = 100; % adjust as needed
mask = bwareaopen(mask, min_size);

% Invert binary mask
mask = ~mask;

% Multiply original image by binary mask to remove black areas
im_clean = uint8(double(im) .* double(repmat(mask, [1,1,size(im,3)])));

% Display original image and cleaned image side by side
subplot(1,2,1);
imshow(im);
title('Original Image');
subplot(1,2,2);
imshow(im_clean);
title('Cleaned Image');

@Rik but this doesn't substract those black areas I am desiring to select. Can you rewrite this so that it does perform what i asked specifically? Thanks!

[Rik]

I cannot do that, because I don't know how to warp the image to remove the discontinuities that would result from removing those blobs.

You keep repeating code. I keep repeating you don't need code yet. First define what you want to do, then start implementing it. You have explained what you want to do.

You are making a start with your definition. You want to measure some sort of density (or packing ratio, etc), and you want to disregard the area with the black blobs when performing this calculation. Would that be a fair description?

[Chanille]

Exactly!!

Answer 1

[Image Analyst]

I believe you need to compute the Euclidean Distance Transform with bwdist and then threshold it. Then call bwareaopen to extract just the biggest regions. Try it.

[Laura]

@Image Analyst Holy guacomole!!! I think I got it!!!! Thank YOU!! @Image Analyst