Digital Signal Processing Reference
In-Depth Information
Appendix C
Hankel Transformation
Hankel transformation is applicable for the image that is circular symmetric.
Hence assume f
ð
x
;
y
Þ
is circular symmetric. The 2D-Fourier transformation in
polar form is given as follows.
F
ð
U
;
V
Þ¼
Z
Z
p
1
g
ð
r
;
h
Þ
e
j2prl
rdrdh
ð
C
:
1
Þ
p
0
where l
¼
Ucos
ð
h
Þþ
V sin
ð
h
Þ
Note that g
ð
r
;
h
Þ
is the polar representation of the
original image f
ð
x
;
y
Þ
(It is also the inverve fourier tranformation of the radon
tranformation of F
ð
U
;
V
Þ
), where U
¼
lcos
ð
h
Þ
and V
¼
lsin
ð
h
Þ
.Asf
ð
x
;
y
Þ
is
circular symmetric, the corresponding radon transformation is identical and hence
g
ð
r
;
h
Þ
is identical for all h. This implies that the radon transformation of F
ð
U
;
V
Þ
is independent of h. (i.e) F
ð
U
;
V
Þ
is circular symmetric and hence it is enough to
compute the values for the co-ordinates on the y-axis. (i.e) U
¼
0 and V
¼
0to
1
.
Particularly for U
¼
0 and V
¼
q, we get l
¼
q sin
ð
h
Þ
. Substituting back to the
Eq. (
C.1
), we get
G
ð
q
Þ¼
Z
Z
p
1
g
ð
r
Þ
e
j2prqsin
ð
h
Þ
rdrdh
ð
C
:
2
Þ
p
0
)
G
ð
q
Þ¼
Z
0
Z
1
g
ð
r
Þ
e
j2prqsin
ð
h
Þ
rdrdh
þ
Z
p
Z
1
g
ð
r
Þ
e
j2prqsin
ð
h
Þ
rdrdh
ð
C
:
3
Þ
p
0
0
0
)
G
ð
q
Þ¼
2
Z
Z
p
1
g
ð
r
Þ
cos
ð
2prqsin
ð
h
ÞÞ
rdrdh
ð
C
:
4
Þ
0
0
G
ð
q
Þ
thus obtained is the hankel transformation of g
ð
r
Þ
: