Obtain 2x2 regions in an image using sliding window

Given an image,I would like to observe 2x2 regions of that image using sliding window.Notice that,as of now I am only interested in obsrving the pixels in that region and not with any further processing.

So its' like I start from the top left,in the 2x2 region there are a set of pixel values,then it goes to the next region and so on till the bottom right.

Answer 1

[Walter Roberson]

If all you are doing is "observing" the pixels, then a double-nested loop is probably easiest; the time taken to "observe" them (display them as an image, or output them to a command line) would be much higher than any inefficiency due to looping.

If you were doing calculations, then there may be faster or more convenient methods.

[Sparsh Garg]

I am currently using this code as reference,but it's not ideal when image is large.For a 37x47 image,it's taking 6 minutes and the process isn't finished yet.

No offense to the original writer of this code,could you give pointers on how I can make it more efficient.

I was thinking of doing nfilter on the image but the function arguemnt is simply impixel

meaning i call nfilter(I,(3,3),impixel)

[Walter Roberson]

nfilter(I,(3,3),impixel)

will never be efficient. The impixel() part of it requires that the user select lines in the image interactively, and that is always going to be slow. And impixel() does not return a function handle, but the third parameter to nlfiter() must be a function handle. Perhaps you are using

nlfilter(I,[3,3],@impixel)

which would pass the 3 x 3 subset to impixel, but impixel would require a user interaction for every window processed.

[Sparsh Garg]

ahhh that sounds reasonable anyways here is the code that I am using,as mentioned earlier would appreciate it if you could give me suggestions on making it faster.

https://in.mathworks.com/matlabcentral/answers/777007-sliding-pixel-window-to-get-8-neighbour-pixel-values

Answer 2

[Image Analyst]

Not sure you've got this solved or not after your last comment, but here is how I'd answer your original question:

rgbImage = imread('peppers.png');
[rows, columns, numColorChannels] = size(rgbImage)
for row = 1 : rows-1
	for col = 1 : columns - 1
		subImage = rgbImage(row:row+1, col:col+1, :);
		imshow(subImage, 'InitialMagnification', 70000, 'XData', [col, col+1], 'YData', [row, row+1]);
		axis('on', 'image');
		pause(0.25);
		caption = sprintf('Showing rows %d-%d, columns %d-%d.\n', row, row+1, col, col+1);
		xlabel(caption);
		fprintf('%s\n', caption);
		drawnow;
	end
end