It really is exactly as the description suggests. In one case the blocks have no overlap so it's like taking a pair of scissors (or something rather more controllable in a straight line!) to your image and cutting it up into pieces that you can put back together to get no more, no less than the entire image.
Blocks with overlap mean that there is an overlap of the image data between neighbouring blocks so that data is repeated. For example, if you used a block size of 4 and an overlap of 2 then your first block would take indices 1:4 in x and y, your 2nd block would take 3:6, your 3rd block would take 5:8, etc. So in this case you will have twice as many blocks as with no overlap as you have a whole extra block between each pair that includes half of one and half of the other. You can have any overlap though, from 1 up to the block size - 1.
