Map a downsampled image to original resolution

I have a small 3D image (500x1000x100) and would like to downsample to 1% of the pixels. I would then like to map this back to a binary (true/false, 1/0) of the same original dimensions, where the relative positions of the downsampled pixels are 1 and everything else is 0.

So maybe like:

small_im = imresize(im,0.01);
%create some sort of mapping
binary_tensor = %true wherever the downsampled image's pixels are present,       
                %relatively speaking. But of the same dims as im

Or this, but it only works if dims are div by 100 and I don't think it works in general:

small_im=im(1:100:end,1:100:end,1:100:end);

mask=false(size(im));
mask(1:100:end,1:100:end,1:100:end)=true;

Think of a 100x100 image. downsample by 99% to 10x10. Take every pixel in that downsample and map it back to the original dimensions and let them be TRUE in a binary tensor of the same dimensions.

I'm trying to recreate a microscope "observing" these pixels in the original sample space, if you were wondering what this is for.

Answer 1

[Walter Roberson]

You can only do this if you used method 'nearest' for the imresize. In all other cases, the value of any output pixels are determined by considering multiple source pixels.

In the case of nearest, you can use linear calculations and floor() to calculate the location of the source pixels.

[Alex G]

Thanks Walter, do you know of any posts that help with the second part. That's what I don't get.

Doing imresize(im, 0.01, 'nearest') is what I'm at since I don't think masking is appropriate for this