Compute Histogram of an image using loops

I'm trying to computer a histogram of an image by using loops. I cannot use the imhist function. How do I do this?

[Matt J]

We can help you with your homework once we see an attempt on your part, and specific questions about where you're stuck.

[Guillaume]

As per Matt's comment, we're not going to do your homework, particularly if you show no effort.

There's not even enough details to answer the question properly. What is the class of the image (double, uint8, uint16, all allowed?). How many bins for the histogram? Is the image colour or greyscale?

Note that the uint8, 256 bins, greyscale image is trivially done in four lines of code including the loop.

[Davidson Bock]

I'm very new to matlab, just started it. The image is supposed to be of type int16 of size 256 x 1. I've done the other 4 part for this homework. I am supposed to use a for loop?

This is proof of my attempt on the homework by completing the other questions

Answer 1

[Image Analyst]

Here's a hint. As you said, you need a loop (one or two depending on how you do it) So you'll have something like this for the 2 loop way:

[rows, columns, numberOfColorChannels] = size(yourImage);
% Preallocate histogram.
h = zeros(1, 256);
for row = 1 : rows
    for col = 1 : columns
        % Now get the gray level
        grayLevel = yourImage(row, col); % Assumes gray scale image.
        % Compute histogram, h.  The index is the gray level for uint8 and the gray level/256 for uint16.  Add 1 to the existing value.
        if isa(class(yourImage), 'uint16'........
            index = ........
        else
        end
        h(index) = ............you finish........
    end
end

I didn't do much other than make the for loops, which you already said you wanted to use. Now you should fill in the inside of the for loop to increment h every time you encounter a pixel of that gray level.

[Davidson Bock]

I found an easier way of just using one loop.

histogram = = zeros(256,1,'int16');
for j = 0 : 255
histogram(j + 1) = number of pixels with intensity j
end

The trouble is that I'm not sure how to find the number of pixels with intensity j, I was thinking of using the find() or sum() function, but I'm still having trouble

[Image Analyst]

No, that's NOT the way to do it with one loop. That inefficient way would require creating a new binary image and then summing it. If you had a megapixel image, that would require a million comparisons and a million sums per gray level. It's far easier to just scan the image and process a pixel at a time:

h = zeros(1, 256);  % Assuming 8 bit image 
for k = 1 : numel(grayImage)
    thisGrayLevel = grayImage(k);
    h(thisGrayLevel + 1) = you finish....
end

Also, don't call your array histogram since that's already the name of a built-in function that you do not want to destroy.

[Davidson Bock]

oh ok, so for that last part, I am guess it will be

 h(thisGrayLevel + 1) = sum(thisGrayLevel(:))

Am I right or close to being right?

[Guillaume]

I'm guessing that Image Analyst's

   h(thisGrayLevel + 1) = ...

was a mistake. The +1 goes on the right hand side of the = and is not added to thisGrayLevel

But frankly, just spending two minutes thinking how to build an histogram by scanning the pixels of an image, or just doing it manually would give you the algorithm. It's extremely simple.

[Image Analyst]

Why do you think it was a mistake? For a gray level of zero, you can't have the zeroeth index, so 1 is added. The first element of h will hold the count for the number of pixels with gray level 0. There will also be a +1 on the right hand side of the equation.

[Guillaume]

Do'h! Of course, you're right. It needs the graylevel+1.

The rest of my comment still stand.