How to sum and plot all pixels of an area in image and plot them?

A new aproach and a NEW question of my previous query.

What I actually want to achieve is to get for every column of the image (see attached image) the sum of all its pixels. But not for the entire image. I want for part inside the red contour.

The counting should start from the bottom of the red segmented contour up until the top red contour. Is there a way to put a marker somewhere in the first one third of the image and tell Matlab to count from that point down to the bottom of the red contour, the total of pixels for every column and plot them.

YOu can see at image A3.jpg, I have marked with green pen the top point and the bottom point, of the area the pixel should be measured.

Is my question clear?

[Adam]

If you have the red surrounding line in Matlab you can just use it as a mask on your original image to remove all the data outside of it, then just sum down each column as you would for a full matrix.

Answer 1

[Image Analyst]

From your other question, you already have the masked image. So to sum vertically, just use sum(). To plot, use plot():

horizontalProfile = sum(maskedImage, 1); % Sum vertically.
plot(horizontalProfile, 'b-', 'LineWidth', 2);
xlabel('Column', 'FontSize', 20);
ylabel('Sum of gray levels in column', 'FontSize', 20);

[Stelios Fanourakis]

Image Analyst, I guess this is the right answer to my query but as you might be able to see, there are a few pixel rows that surpass the red top boundary. If you notice at the right end side of the picture there is the ruler for all ultrasound units. Slightly above the red contour starts the first dash line of the ruler. I want the sum of all pixels to include those rows of pixels above the red contour. So I need to slightly modify it and add around 3-4 pixel rows

[Stelios Fanourakis]

In addition, when we say 0.5 thickness of a line is it data units or pixels? I found this in the drawline() settings in the figure when we display the image.

[Stelios Fanourakis]

Also, I need a way to get the difference out of two plots. Assumingly, we have two plots from two different masks. I need to find the differences between those two images.

[Image Analyst]

Why cannot you get the images before someone has marked them up with a red line?

[Stelios Fanourakis]

Because the segmentation contour is essential. At least to the point of the soft tissue (bottom red line).

Image Analyst, I am looking for a way to define a specific pixel as my starting point in the image, where the sum of pixels will begin. So I need actually, a method to allow me define some points of image as a starting and end markers. Define a specific area to sum all pixels of it. Then, I will subtract some other pixels that are not needed in the calculation.

I'll come back again.

[Stelios Fanourakis]

Image Analyst, how can I say start summing all pixels from the point (136,123) on image. Scan vertically down for all columns until the point (548,125). The scanning will sum all pixels until it reaches the first red pixel which is the bottom red line and then stops. I want this calculations. If possible, I want for every column the number of pixels. Any suggestion please?

Thanks in advance.

[Stelios Fanourakis]

I use this code but I get the error "Assignment has more non-singleton rhs dimensions than non-singleton subscripts" at line s(i,j)=sum(s) .

Remember, I want to sum all pixels of a certain area of the image and the summation will go on until it finds the first red pixel where it should stop and move to the next column. Please help

clc;
s = imread('A2.jpg')

for i = 136:548
    for j = 123:length(s)
        s(i,j)=sum(s)
        if s(i,j)== 100
            break
        end
        s = s+1;
        end
        end
        

[Image Analyst]

You have the masked image already from my other answer. If you want to start summing from y = 123 to the bottom of the mask, just do this

row1 = 123; % Whatever...
s = zeros(1, columns); % Initialize sums for every column.
for col = 1 : columns
    row2 = find(mask(:, col), 1, 'last'); % Get bottom row of mask
    if ~isempty(row2)
        % Get the sum from row1 to row2
        s(col)  = sum(grayImage(row1:row2, col));
    end
end

Answer 2

[Walter Roberson]

You can threshhold on bright white. Suppose you do that getting threshimg that is 1 where the bright white is. Then

counts = sum(cumprod(~threshimg))

will be the per-column count of pixels from the top to the first bright white pixel in the column.

The sum of the cumprod of the negated image is a clever vectorized way of doing the equivalent of

nr = size(threshimg,1); nc = size(threshimg,2);
for K = 1 : nc
    counts(K) = find([threshimg(:,K);true], 1) - 1;  %force artificial boundary in case no line
end

[Stelios Fanourakis]

@Walter. I'd better have a margin of two points for every column. From top red pixel to bottom red pixel and count all the pixels for every column. It is important for the summation of pixels to end where the red dots of the contour are. In the case of the top red boundary I'd like to add 1-2 more pixel rows since if you notice the gray area of the ultrasound image does not end where the top red boundary lies. It is exceeded by a few pixels on top of the red boundary. I will also need to include those few pixels. ANy idea?

[Walter Roberson]

WIll the red be actually present in the image, or did you just show us it for the purpose of explanation ?

It is true that I was not taking into account the annotations around the obvious image area, but I guess those annotations are part of the images themselves, so the algorithm would need to be improved to find the location of the inner image.

[Image Analyst]

No, he already had another, earlier question were we solved the problem of masking out everything except the stuff inside the red outline.

[Stelios Fanourakis]

Image Analyst, how can I say start summing all pixels from the point (136,123) on image. Scan vertically down for all columns until the point (548,125). The scanning will sum all pixels until it reaches the first red pixel which is the bottom red line and then stops. I want this calculations. If possible, I want for every column the number of pixels. Any suggestion please?

Thanks in advance.

[Stelios Fanourakis]

I use this code but I get the error "Assignment has more non-singleton rhs dimensions than non-singleton subscripts" at line s(i,j)=sum(s) .

Remember, I want to sum all pixels of a certain area of the image and the summation will go on until it finds the first red pixel where it should stop and move to the next column. Please help

clc;
s = imread('A2.jpg')

for i = 136:548
    for j = 123:length(s)
        s(i,j)=sum(s)
        if s(i,j)== 100
            break
        end
        s = s+1;
        end
        end
        

[Walter Roberson]

Your s is 3D, since it is an RGB image. sum(s) is going to be 3D as well, since the default summation dimension is the first non-singular dimension, which is going to be the first dimension,leaving you with 1 x columns x 3 in size. You cannot store that into a scalar location s(i,j) .

Watch out: length(s) means

temp = size(s);
if any(temp == 0)
    s_length = 0;
else
    s_length = max(temp);
end

It does not mean the number of rows or columns in s.

If, somehow, you were able to store the result of the sum() into the scalar location, then you would be modifying progressively modifying the same array that you are using sum() over, and you would be adding 1 to each location as well. The code just does not make any sense.