How to find high contrast objects in image?

I have matrix 50x200 and i want script that will go through the matrix and compare values with its surrounding values which gives contrast ratio, very higher results will mean object with higher contrast. Usage is detecting objects in image from luminance analyser detected while driving in car.

[Jakub]

Each value in matrix is luminance value of each image`s pixel.

Answer 1

[TED MOSBY]

Hi Jakub,

The contrast ratio is the absolute difference between the center pixel and the mean of its 8 neighbors divided by the neighbor mean. You can simply loop through the pixels and extract a 3*3 block around each of it, then calculate the contrast.

I have written an example script for your reference:

imageMatrix = rand(50,200);  % This is just an example matrix
[m, n] = size(imageMatrix);

contrastMatrix = zeros(m, n);

% Loop through the matrix excluding the boundary pixels
for i = 2:m-1
    for j = 2:n-1
        % Extract the 3x3 neighborhood centered at (i,j)
        neighborhood = imageMatrix(i-1:i+1, j-1:j+1);
        
        % Calculate the sum of the 3x3 block and subtract the center pixel
        centerPixel = imageMatrix(i,j);
        localMean = (sum(neighborhood(:)) - centerPixel) / 8;
        
        % Compute the contrast ratio
        if localMean ~= 0
            contrastMatrix(i,j) = abs(centerPixel - localMean) / localMean;
        else
            contrastMatrix(i,j) = 0;
        end
    end
end

% Visualize the contrast matrix
figure;
imagesc(contrastMatrix);
colorbar;
title('Contrast Ratio Map');
xlabel('Pixel Column');
ylabel('Pixel Row');

The output appears as below:

Hope this helps!

[Walter Roberson]

This sounds like a job for

abs(conv2(double(imageMatrix), [1 1 1; 1 -1 1; 1 1 1]/8, 'same'))