How to vectorize moving window across an image?

Hello,

I am trying to learn faster image processing techniques. I currently have implemented a double for loop that calculates a z-score and "throw away" the value if it is too close to the mean.

One option I considered was using a par for loop, but from my understanding, implementation using vectorization is a faster/better practice.

Here is the snippet of what I am trying to do:

inFrame = imread('Lenna.png');

% Convert frame into grayscale, use standard NTSC Conversion
grayFrame = 0.2989*inFrame(:,:,1) + 0.5870*inFrame(:,:,2) + 0.1140*inFrame(:,:,3);
[height, width, colormap] = size(grayFrame);

% Preallocate for window matrix
windowSize = 5;
window = zeros(windowSize);
zScoreMatrix = zeros(size(grayFrame));

% Begin moving window across image

 for i = 3:height-2
     for j = 3:width-2
        window = grayFrame(i-2:i+2,j-2:j+2);
        window = double(window);

        mean = mean2(window);
        std_dev = std2(window);
        zScore = (window(3,3) - mean)/std_dev;
        
        if abs(window(3,3) - mean) < 3 
            zScoreMatrix(i,j) = 0;
        else
            zScoreMatrix(i,j) = zScore;
        end
    end
 end

Answer 1

[Image Analyst]

Try using imfilter and stdfilt

% Demo by Image Analyst
% Initialization Steps.
clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 18;
markerSize = 40;

grayFrame = imread('cameraman.tif');
% Display the image.
subplot(2, 2, 1);
imshow(grayFrame, []);
impixelinfo;
title('Original Image', 'FontSize', fontSize, 'Interpreter', 'None');
axis('on', 'image');

% Blur the image.
windowSize = 5;
kernel = ones(windowSize) / windowSize^2;
blurredImage = double(imfilter(grayFrame, kernel, "replicate"));
% Display the image.
subplot(2, 2, 2);
imshow(blurredImage, []);
impixelinfo;
title('Blurred Image', 'FontSize', fontSize, 'Interpreter', 'None');
axis('on', 'image');

% Get the standard deviation of the image.
stdDevImage = stdfilt(grayFrame, ones(windowSize));
% Display the image.
subplot(2, 2, 3);
imshow(stdDevImage, []);
impixelinfo;
title('Standard Deviation Image', 'FontSize', fontSize, 'Interpreter', 'None');
axis('on', 'image');

% Compute the z score.
zScoreMatrix = (double(grayFrame) - blurredImage) ./ stdDevImage;
% Display the image.
subplot(2, 2, 4);
imshow(zScoreMatrix, []);
impixelinfo;
title('Z Image', 'FontSize', fontSize, 'Interpreter', 'None');
axis('on', 'image');

[Sonicflash]

I followed through your demo, and it worked great. I didn't know of impixelinfo before, either.

Thank you kindly. Really appreicate it.