Currently the comparison algorithm requires compared images to be exactly the same size because it uses a dumb pixel-by-pixel comparison with a tolerance to identify differences between two images.
I'd like the ability to compare images of different sizes for variations in color and shape. For this we'd need a fuzzy image comparison algorithm that can determine if variations between two pictures have, overall, the same shape. This can likely be done by checking gradients between to two images to identify the rate of change is similar between two images rather than doing a direct color comparison on a pixel-by-pixel basis.
Currently the comparison algorithm requires compared images to be exactly the same size because it uses a dumb pixel-by-pixel comparison with a tolerance to identify differences between two images.
I'd like the ability to compare images of different sizes for variations in color and shape. For this we'd need a fuzzy image comparison algorithm that can determine if variations between two pictures have, overall, the same shape. This can likely be done by checking gradients between to two images to identify the rate of change is similar between two images rather than doing a direct color comparison on a pixel-by-pixel basis.