how do save them all up and make a table with the values from all of my images?

my code has a table wth only one row in it and all 3 columns the same??? How do I save them all up and make a table with the values from all my images? Thank you!

i also have an error:

Unrecognized function or variable 'meanOfPixelsInRoi'.

Error in line 32

info_table = table(meanOfPixelsInRoi, mean(grayImage(roi)), meanOfPixelsInRoi,'VariableNames', {'MeanOfPixelBlobs1', 'MeanOfPixelBlobs2', 'MeanOfPixelBlobs3'})

% Loads the z, 32 bit, single channel image 
rgbImage = imread('test1.jpg');
  
grayImage = double(rgbImage(:, :, 2));


%replace each pixel with the average of its 3x3 neighbors with 
%filtering with a radius of 2 pixels

mean3=conv2(grayImage,[0 1 0; 1 1 1; 0 1 0]/5,'same');
imshow(mean3, []);

%set a pixel threshold cutoff in imagej it was 60 - minimize background by choosing a 
%value cutoff, such that every pixel less than that value is considered one class, 
%while every pixel greater than that value is considered the other class.


roi = grayImage < 60; % Define roi by thresholding

%convert the intensity values of the background to "nothing" 
%rather than to 0 and keeps the other values


%divide this image by itself -> ROI

roifinal = roi./roi;
imshow(roifinal, []);


% Tabulate individual mean values 
meanOfPixelsInRoi = mean(grayImage(roi));
info_table = table(meanOfPixelsInRoi,  mean(grayImage(roi)), meanOfPixelsInRoi,'VariableNames', {'MeanOfPixelBlobs1', 'MeanOfPixelBlobs2', 'MeanOfPixelBlobs3'})
finalcount = sum(meanOfPixelsInRoi) 
disp(info_table)

[Walter Roberson]

I substituted a different image, and there was no problem executing the code.

[Neo]

Can I see your result @Walter Roberson

[Walter Roberson]

filename = 'flamingos.jpg';
% Loads the z, 32 bit, single channel image 
rgbImage = imread(filename);
  
grayImage = double(rgbImage(:, :, 2));


%replace each pixel with the average of its 3x3 neighbors with 
%filtering with a radius of 2 pixels

mean3=conv2(grayImage,[0 1 0; 1 1 1; 0 1 0]/5,'same');
imshow(mean3, []);

%set a pixel threshold cutoff in imagej it was 60 - minimize background by choosing a 
%value cutoff, such that every pixel less than that value is considered one class, 
%while every pixel greater than that value is considered the other class.


roi = grayImage < 60; % Define roi by thresholding

%convert the intensity values of the background to "nothing" 
%rather than to 0 and keeps the other values


%divide this image by itself -> ROI

roifinal = roi./roi;
imshow(roifinal, []);

% Tabulate individual mean values 
meanOfPixelsInRoi = mean(grayImage(roi));
info_table = table(meanOfPixelsInRoi,  mean(grayImage(roi)), meanOfPixelsInRoi,'VariableNames', {'MeanOfPixelBlobs1', 'MeanOfPixelBlobs2', 'MeanOfPixelBlobs3'})

info_table = 1×3 table
    MeanOfPixelBlobs1    MeanOfPixelBlobs2    MeanOfPixelBlobs3
    _________________    _________________    _________________

         41.457               41.457               41.457      

finalcount = sum(meanOfPixelsInRoi) 

finalcount = 41.4573

disp(info_table)

    MeanOfPixelBlobs1    MeanOfPixelBlobs2    MeanOfPixelBlobs3
    _________________    _________________    _________________

         41.457               41.457               41.457      

[Image Analyst]

See, like I said somewhere (your other post maybe), all three values are identical.

[Neo]

To help me see the blobs better can you insert a line that draws around the blobs of pixels being analysized?

[Neo]

Thank you both @image analyst and @walter Robertson

[Image Analyst]

See this snippet and adapt as needed.

% Plot the borders of all the blobs in the overlay above the original grayscale image 
% using the coordinates returned by bwboundaries().
% bwboundaries() returns a cell array, where each cell contains the row/column coordinates for an object in the image.
imshow(originalImage); % Optional : show the original image again.  Or you can leave the binary image showing if you want.
% Here is where we actually get the boundaries for each blob.
boundaries = bwboundaries(mask);
% boundaries is a cell array - one cell for each blob.
% In each cell is an N-by-2 list of coordinates in a (row, column) format.  Note: NOT (x,y).
% Column 1 is rows, or y.    Column 2 is columns, or x.
numberOfBoundaries = size(boundaries, 1); % Count the boundaries so we can use it in our for loop
% Here is where we actually plot the boundaries of each blob in the overlay.
hold on; % Don't let boundaries blow away the displayed image.
for k = 1 : numberOfBoundaries
    thisBoundary = boundaries{k}; % Get boundary for this specific blob.
    x = thisBoundary(:,2); % Column 2 is the columns, which is x.
    y = thisBoundary(:,1); % Column 1 is the rows, which is y.
    plot(x, y, 'r-', 'LineWidth', 2); % Plot boundary in red.
end
hold off;
caption = sprintf('%d Outlines, from bwboundaries()', numberOfBoundaries); 
fontSize = 15;
title(caption, 'FontSize', fontSize); 
axis('on', 'image'); % Make sure image is not artificially stretched because of screen's aspect ratio.

Answer 1

[Image Analyst]

You need to get all your filenames in a cell array (see the FAQ)

Then

You'd need to do something like this inside your loop over k:

for k = 1 : numImages
    % Loads the z, 32 bit, single channel image 
    rgbImage = imread(allFileNames{k});
    if size(rgbImage, 3) > 1
        % Take green channel.
        grayImage = double(rgbImage(:, :, 2));    
    else
        % It's already gray scale.
        grayImage = double(rgbImage);    
    end
    
    %replace each pixel with the average of its 3x3 neighbors with 
    %filtering with a radius of 2 pixels
    
    mean3=conv2(grayImage,[0 1 0; 1 1 1; 0 1 0]/5, 'same');
    imshow(mean3, []);
    drawnow; % Force it to repaint NOW.
    
    %set a pixel threshold cutoff in imagej it was 60 - minimize background by choosing a 
    %value cutoff, such that every pixel less than that value is considered one class, 
    %while every pixel greater than that value is considered the other class.    
    
    roi = grayImage < 60; % Define roi by thresholding
    
    %convert the intensity values of the background to "nothing" 
    %rather than to 0 and keeps the other values    
    
    %divide this image by itself -> ROI
    % The next line is not needed at all.
    roifinal = roi./roi;  % What the heck?  This will give just an image of 1s and nans.
    imshow(roifinal, []);    
    
    % Tabulate individual mean values 
    meanOfPixelsInRoi(k) = mean(grayImage(roi));
end % of loop over k for all images.

% Then after the loop over all images, create your table
info_table = table(allFileNames, meanOfPixelsInRoi(:), 'VariableNames', {'FileName', 'Mean of pixel blobs'});
finalcount = sum(meanOfPixelsInRoi) 
disp(info_table)

Make sure allFileNames is a column vector where the filenames are listed vertically. To make sure you can do this

allFileNames = reshape(allFileNames, [], 1);

[Neo]

@Image Analyst

because I set the threshold does not mean it is now the ROI (according to the definition I am trying to achieve).

My criteria for the ROI:

After I set the threshold I want to set the other pixels that did not meet the threshold to nan.

(Which is why I want to incorporate the nan line)

Then I want to divide that image by itself, which would give me the ROI.

since what I have would not work can you suggest what will?

Thank you.

[Image Analyst]

Your ROI is defined as pixels with value less than 60. You don't need roiFinal. Just delete all lines that mention roiFinal and you'll get the right results.

[Neo]

@Image Analyst

I greatly respect your opinion Analyst. however as you mentioned in another post, my roi is what I define it. I would like todefine it by 1. thresholidng the pixels of the image after the line with mean3 then, 2. divide that image by itself.

If what I have will not do it I am sure you can explain what I could do that will. Whayt do you suggest, my Lord?

Thank you!

Kind wishes,

Neo

[Image Analyst]

I'm sorry but I really don't know how to respond. When you say that you want a binary image (mask) of all pixels with gray levels less than 60, then that mask defines the ROI. Again you can simply get the mean value of all those gray levels less than 60 by doing

mask = grayImage < 60; % The variable "mask" is the ROI and is a logical (binary) image
meanLessThan60 = mean(grayImage(mask));

There is no need to divide the image by itself. There just isn't. And there is no need to use nans at all. Again, it's just simply not necessary.

[Neo]

What do you suppose that I would get if I did divide the image by itself? I only suggest it because when I found theROI by hand I had to do that.

[Image Analyst]

I would expect that you'd either get 1, or if the denominator was 0 you'd get either NaN or Inf. Let's try it and see:

u1 = logical(1);
u0 = logical(0);
r = u0 / u0

r = NaN

r = u0 / u1

r = 0

r = u1 / u0

r = Inf

r = u1 / u1

r = 1

Again, there is no need to create an image with those values. I don't care if you have to do it "by hand" or programmatically in your script. There is no need to do it at all in order to get what you want to get. It's just an intermediate step that is totally not needed.

[Neo]

What I want to get ultimately in the end are the means of these blobs that would give me consistent and meaningful data. For example, this image that i have circled are the blobs that I would get the mean of.

[Neo]

And I’m trying to visualize how there is no difference. If we think about it theoretically: if I have an image with pixel values less than a threshold 60 set the other pixels to nan then divide that image by itself shouldn’t it give me just the pixels that I’m interested in? If not okay I’ll leave it alone and will exclude it from my future analysis.

[Image Analyst]

OK you have done some kind of segmentation that gave you more blobs than you want. How was that segmentation done? Now you need to improve or continue the segmentation so that you get only the red-circled blobs. What is different, either in the binary image or in the original gray scale image, between the red-circled regions and the uncircled regions?

[Neo]

Sorry. I’ve done the following filtering:

1. Threshold

2. Nan

3. Divide by itself

4. Multiply result by original image

I want most of the blobs (excluding the ones blended together) I just circled a few as n example. The difference is that I’ve now filtered out the image that is not interesting or won’t give me valueable data.

[Image Analyst]

You get a mask by thresholding.

mask = grayImage < 60;

If you want you can use the mask, by using it as an index

grayImage(~mask) = 0; % Erase/blacken outside the mask.

or by multiplying

grayImage = grayImage .* uint8(mask);

to erase or black the gray scale image outside the mask, though this is really not necessary.

So I guess all the white regions in your mask are where the original image values are less than 60. But again, tell me what is special about the red-circled regions. Are they less than 10? Are they smoother or rougher than the other regions? What's different about them? I can't tell just by looking at the binary image.

Do you just want to manually trace them with drawpolygon()? (Demo attached.)

Attach the original image that you thresholded. Maybe seeing that will help me decipher better what you're trying to do.

[Neo]

I will use a new image:

tiffex is original image

tiffex1 is the image after I did the equivalent of the mean3 convolution

tiffex2 is the image after the threshold is applied

tiffex3 is the image after I have divided it by itself (roi)

tiffexfinal is the image after multiplying the original image by the roi and inc the brightness of the image (ultimatley the image that I wanted).

I am interested in all of the red blobs remaining.

Here are the values of the blobs for tiffexfinal:

area mean min max

715.192 276.919 20 1041

188.611 125.972 36 429

1926.805 364.926 22 1046

2608.487 558.045 25 1500

here are the values of the blobs after thresholding (tiffex2)

715.192 270.560 60 707

188.611 108.010 60 182

1926.805 354.892 60 763

2601.306 547.864 60 1297

although the mean and area are the same for both image (ignore the numerical difference) the min and max are drastically different which I think excludes data that I might want. Which is why I do the other steps. Please let me know what you think my Lords! hahaha

[Neo]

@Image Analyst & @Walter Roberson

[Image Analyst]

So can you verify that for those red blobs

you want to find the area, mean intensity of red, max intensity of red, and min intensity of red? If so, I can show you the best way to get those measurements.

[Neo]

Yeah. But I think I have done those measurements the best way, right?

[Neo]

Actually I just want the the area and mean of the blobs but I think that the min shows some blobs are in a certain cut off. The cut off May be important

[Image Analyst]

What do you mean by "cut off"?

[Walter Roberson]

You threshold at 60. Of course they are cut off.

[Neo]

Yes but if I divide the image by itself and the other steps I mentioned then the cut off will be lower. So maybe the other steps are important? The cut off as in the min values after thresholding.

[Image Analyst]

Again, what do you mean by "cut off"? Please, please use another, more descriptive word. Is the cutoff the threshold value? Or do you mean that the binary image blobs are smaller/cutoff/truncated because an improper threshold was chosen?

Try the attached program.

The program gives the area, mean, min, and max intensities and doesn't do any of that nonsense with NaNs or dividing the image by itself.

[Neo]

@Image Analyst something close to the latter of what you have written.

notice that min (what i meant by cut off) is all 60 after just threholding the pixels in tiffex2 image but after i have done all of the other steps which we are considering to be unnecessary the min values are now below 60 (in the tiffexfinal image). I think this means that if i did not do the other "unnecessary" steps that I would be restrciting the values to 60 rather than other values that may exist that are less than 60.

What do you think my Lords? @Image Analyst @Walter Roberson

[Neo]

@Image Analyst Does it give the mean of the red regions highlighted?

i would like to have those tabulated similar to my code but it looks something like what I am interested in!

[Image Analyst]

Not sure I follow you so I'll try to explain. The image was thresholded at 60 so the detected foreground blobs are composed of pixels that have gray levels of 60 or higher. Any of the stats I compute with regionprops are done on those blobs only so they will have a min no less than 60. Their max and mean will of course be higher than 60. There are also pixels still in the image that have gray levels less than 60 but those are background, non-roi pixels whose gray levels are not used in any of the computations about area and intensity.

You say "I think this means that if i did not do the other "unnecessary" steps that I would be restrciting the values to 60 rather than other values that may exist that are less than 60." Not sure if that means you WANT values less than 60. You say if you didn't do that unnecessary stuff you'd have ONLY values less than 60. That is not what I did. I computed using only values 60 or greater. If you want 59 or less you'd do

mask = grayImage < 60;

and then regionprops(), and in that case you'd get different blobs and different areas and intensity measurements since the measurements would all be made on the dark regions.

[Neo]

By the dark regions you mean the background? I’m not interested in that. I’m only interested in the red blobs.

[Image Analyst]

Please explain why you said

"I think this means that if i did not do the other "unnecessary" steps that I would be restrciting the values to 60 rather than other values that may exist that are less than 60."

Let's try to examine the sentence.

"I think this means that if i did not do the other "unnecessary" steps" <= OK that means not doing the "dividing the image by itself" step.

"I would be restrciting the values to 60" <= No idea what you mean here. In my code I did not use exactly and ONLY 60. I used a range of 60 to 255 for the foreground and 0-59 for the dark background. Exactly what do you mean by "restricting" and why is it restricted to only one gray level?

"rather than other values that may exist that are less than 60." <= to me this implies that you ARE interested in the values less than 60 but my code ignores those values and you don't like that fact.

And actually my code does ignore the dark background and does only analyze the bright red blobs. So why is that not what you want? You seem to say it is what you want but you don't like or understand my solution for some reason. That seems contradictory to me. Please elaborate.

[Neo]

@Image Analyst

My apologizes you are quite right. Your code does indeed do what I want as well. Thank you.

what I meant by my sentences was that if you look at the min values for tiffex2 you will see that all of the mins are 60 but in tifexfinal (where I have done additional steps) you will see that the mins are less than 60 and various.

I only want analyze the red blobs as your code does, but I did not know if stopping at the tiffex2 step (only thresholding no dividing or nans) would restrict the mean and area values (i dont know, does it?). I suppose it may since we see a difference in the numerical value of the min in tiffex2 vs tiffexfinal min but the means are the same so maybe it doesnt.

I noticed that your code does not produce the individual tabulated mean values of the blobs. Why is that?

[Image Analyst]

Well, tiffex2 is created in your code, not mine, and like I said, I don't think it's necessary.

If you want the data to come back as a table, pass in 'table' as the first argument to regionprops(). I usually get the results back as a structure array because that's the way they've done it historically and I'm used to that way. To get back as table

props = regionprops('table', mask, grayImage);

To get back as a structure array

props = regionprops(mask, grayImage);
% To get areas in an array from the structure array
allAreas = [props.Area]

If it's doing what you want now, then please click the "Accept this answer" link. Thanks in advance.