Hi I have a problem when I preform the convolution original image with kernel of ones as the size of the kernel increases the output is appearing totally white Image.

            Image = imread('cameraman.tif');
            Image1 = double(Image);

For i=1:length(Image1)
  kernel=ones(i);   K=Conv2(Image1,kernel,’same’);     Imshow(K)     
end

Answer 1

[Walter Roberson]

            Image = imread('cameraman.tif');

Image ranges from uint8 0 to uint8 255

            Image1 = double(Image);

Image1 ranges from double 0 to double 255.

             kernel=ones(i);   K=Conv2(Image1,kernel,’same’);

Because of the convolution, up to i^2 locations are added together. The output could be between double 0 and double 255*(i^2) . When you add together numbers that are integral values, then you get results that are integral values, so you are going to get results that are strictly 0, 1, 2, ...

             Imshow(K)

K is double 0 to (possibly) double 255*(i^2).

When imshow() is passed a double precision array, then the data is assumed to be between double 0 and double 1, and everything else is clipped to that range. But the values happen to be strictly 0, 1, 2, ... and clipping that to the range 0 to 1 will result in 0 where the data was 0, and will result in 1 everywhere else.

[Walter Roberson]

Image = imread('cameraman.tif');
Image1 = im2double(Image);

tiledlayout('flow');
for i=5:5:40
  kernel = ones(i) / (i^2);
  K = conv2(Image1,kernel, 'same');
  nexttile();
  imshow(K); title(string(i));
end