The image size of the destination image was used to determine if a source
image was positioned outside the destination image, that no intersection
could occur.  Actually for these two cases the size of the source image
has to be used!
Signed-off-by: Stefano Sabatini <stefasab@gmail.com>