
6.4 Smoothing
255
If we only have a single noisy picture available, we can attempt to reduce the
noise level by
local
averaging, i.e., we give each point a new gray level which is
the average of the original gray levels in some neighborhood of the point
(including the point itself). This is relatively straightforward if the noise is
finer grained than the smallest picture detail of interest; it should then be pos-
sible,
by averaging over a small neighborhood of every point, to reduce the
noise while keeping the blur at a negligible, or at least tolerable, level. Alter-
natively, we may be able to remove the noise by suppressing or deemphasizin ...