how exactly does applying the filter shift the input to the left by one?
motoole2
Ah, looks like I missed up slightly on this slide after all. The result of a convolution operation would cause the image to shift right by one; the result of a correlation operation would cause the image to shift left by one.
In either case, starting from the equations on slide 55, we can simplify the expression because the kernel is non-zero for only one entry. This results in the following (for a convolution):
how exactly does applying the filter shift the input to the left by one?
Ah, looks like I missed up slightly on this slide after all. The result of a convolution operation would cause the image to shift right by one; the result of a correlation operation would cause the image to shift left by one.
In either case, starting from the equations on slide 55, we can simplify the expression because the kernel is non-zero for only one entry. This results in the following (for a convolution):
$$(f * I)(x,y) = I(x-1,y)$$
where $I$ is the image and $f$ is the filter.