Image gradients and Convolution mask

For digital image, the gradients of 2D functions can be written as the following:

$\begin{align*}G_x &\simeq f[i, j+1] - f[i, j]\\G_y &\simeq f[i+1, j] - f[i, j],\end{align*}$

noticing that $j$ corresponds to the $x$ direction of the digital image, while $i$ corresponds to the $y$ direction of the digital image.

Published by da