Digital Signal Processing Reference
In-Depth Information
H
:
z
(
X
)
=
b
(
X
)
+
n
(
X
)
(6)
0
i
i
i
2.2
Top-Hat Filter
The background of IR image sequence is usually extensive and changes gently, ac-
counting for the low-frequency part of the image spatial frequency spectrum, while
the target is small and isolated like a bright point, accounting for the high-frequency
part of the image spatial frequency spectrum. Thus high-pass filters are proper useful
for preprocessing. As one of the high-pass filters, Top-hat is used to eliminate the
element changes indistinctively and detect the element of high-frequency by morpho-
logical operation, as shown in equation (7).
Hat
(
f
)
=
f
−
(
f
b
)
(7)
Here
f
stands for the grayscale image,
b
is the structure element, and
means to do open operation on image
f
by the structure element
b
.
Figure 4 shows the original image and the preprocessing result in the form of
grayscale image and histogram. The result indicates that Top-hat operator does well in
burring and eliminating background noise like clouds.
(
f
b
)
(a) (b)
(c) (d)
Fig. 4.
Preprocessing by Top-hat: (a) The original IR image in the sky scene; (b) Histogram of
the original grayscale image; (c) Preprocessing result by Top-hat filter; (d) Histogram of the
preprocessing result
