Hi,
I don't understand why the second statement is true. Because, according to the course, to speed the edge detection, "image derivatives are computed by convolving
with the derivative of a Gaussian" (and not by the derivative of the signal f!)
Thanks,
Loïc