drying to scale down image by factor of 2 using loops

Hi, Ive been stuck trying to figure out how to scale down a greyscale image using loops and the mean function. This is what I am doing so far which I know is incorrect but Im just really struggling with getting the values from the 2D array input and seperating it into multiple smaller arrays to then take the mean of them.

function [HalfmyIm] = HalveImage(myIm)
%UNTITLED2 Summary of this function goes here
%   Detailed explanation goes here
%myIm=uint8(myIm);
[rows, cols]=size(myIm);
mean_values = zeros(2,2);
for i= 1:rows
    for j= 1:cols 
        % Half(i,j)=myIm(i:i+1, j:j+1)
        % mean(Half(i,j))
        r_idx = [i i+1]
        c_idx = [j j+1]
        smallA= myIm(r_idx, c_idx);
        mean_values(i, j) = mean(smallA(:));
        
    end
    
end
disp(mean_values);
end

Answer 1

[DGM]

I'm going to generalize a little bit. Scaling factor can be set as an internal parameter. The code supports any number of image channels. This can be turned into a function as needed.

inpict = imread('peppers.png');

% the main parameter
rsfactor = 4; % i'm using 4 so the change is noticeable on the forum

% the last output of size() is a product term, not strictly a size
[rows, cols, chans, ~] = size(inpict);

% make sure the size is compatible with the scaling factor
if any(mod([rows cols],rsfactor))
    error('HALVEIMAGE: page geometry must be an integer multiple of %d',rsfactor)
end

% preallocate output image to the same class as the input
outpict = zeros(rows/rsfactor,cols/rsfactor,chans,class(inpict));

% increment in steps of rsfactor
for i = rsfactor:rsfactor:rows
    for j = rsfactor:rsfactor:cols
        % get sample region
        r_idx = (i-rsfactor+1):i;
        c_idx = (j-rsfactor+1):j;
        smallA = inpict(r_idx, c_idx, :); % all channels
        
        % average on first two dimensions only
        outpict(i/rsfactor, j/rsfactor, :) = mean(smallA,1:2);
    end
end

size(inpict)

ans = 1×3

384 512 3

size(outpict)

ans = 1×3

96 128 3

imshow(outpict)