Implement average filter without using built-in functions

I am trying to blur the image, but I did not succeed. I am keep getting almost black image? What am I missing here?

clc;
clear all;

img = imread("Q3_Input", "tif");
imshow(img);
[M, N] = size(img);
filter = averageFilter(img, M, N);
%blurredImage = conv2(single(img), filter, 'full');
figure;
imshow(filter, []);

%{
    since the filter 3x3

    i == 1 & j == 1 or
    i == 1 & (N - j) == 0 or
    (M - i) & j == 1 or
    (M - i) & (N - j) == 0
    covers the corner areas 
    
    x = covered areas
    __________
    |_x_|__|_x_|
    |__|__|__|
    |_x_|__|_x_|
    
    the other coverts the middle of the area

    |__|_x_|__|
    |_x_|_x_|_x_|
    |__|_x_|__|
%}
function img = averageFilter(image, M, N)
    newImg = zeros(M, N);
    for i = 1: M
        for j = 1: N
            if i == 1 
                if j == 1
                    summation = 0;
                    for k = i: i + 1
                        for l = j: j + 1
                            summation = summation + image(k,l);
                        end
                    end
                    newImg(i, j) = ceil(summation / 4.0);
                elseif (N - j) == 0
                    for k = i: i + 1
                        for l = j - 1: j
                            summation = summation + image(k,l);
                        end
                    end
                    newImg(i, j) = ceil(summation / 4.0);
                else
                    for k = i: i + 1
                        for l = j - 1: j + 1
                            summation = summation + image(k,l);
                        end
                    end
                        newImg(i, j) = ceil(summation / 6.0);
                end
            elseif (M - i) == 0
                if j == 1
                    for k = i - 1: i
                        for l = j: j + 1
                            summation = summation + image(k,l);
                        end
                    end
                    newImg(i, j) = ceil(summation / 4.0);
                elseif (N - j) == 0
                    for k = i -1: i
                        for l = j -1: j
                            summation = summation + image(k,l);
                        end
                    end
                    newImg(i, j) = ceil(summation / 4.0);
                else
                    for k = i - 1: i
                        for l = j - 1: j + 1
                            summation = summation + image(k,l);
                        end
                    end
                    newImg(i, j) = ceil(summation / 6.0);
                end
            else
                if j == 1
                    for k = i - 1: i + 1
                        for l = j: j + 1
                            summation = summation + image(k,l);
                        end
                    end
                    newImg(i, j) = ceil(summation / 6.0);
                elseif (N - j) == 0
                    for k = i - 1: i + 1
                        for l = j - 1: j
                            summation = summation + image(k,l);
                        end
                    end
                    newImg(i, j) = ceil(summation / 6.0);
                else
                    for k = i - 1: i + 1
                        for l = j - 1: j + 1
                            summation = summation + image(k,l);
                        end
                    end
                    newImg(i, j) = ceil(summation / 9.0);
                end
            end
        end
    end
    img = newImg;
end

original image

Answer 1

[Matt J]

In all likelihood, you have not converted your image to floating point

img = im2double(  imread("Q3_Input", "tif") );

[kaan köken]

If I get rid of the floating points, and do not change the im read. Also, If I add this I am getting white image as result.

blurredImage = img .* filter;
figure;
imshow(blurredImage, []);

However, if I change the imread to your version I am getting this image. Can I do it without floating point solution?

[Image Analyst]

No. Computing the mean in a window inherently gives you a floating point number. However, if you want, you can cast the final floating point image into uint8 after the whole window scanning process has finished.