Information Technology Reference
In-Depth Information
1
r
c
r
c
∑∑
∑∑
DI
=
Re
f
(
i
,
j
)
−
fR
(
i
,
j
)
(14)
r
×
c
i
==
11
j
i
==
11
j
Ref is the reference image, r and c represent the image's width and length. Edge
preservation (EP) expresses a image's edge information. It and average energy of
wavelet coefficients (WAE) both can indicate a image's resolution and clarity.
r
c
[
]
∑∑
Q
(
i
,
j
)
⋅
w
(
i
,
j
)
+
Q
(
i
,
j
)
⋅
w
(
i
,
j
)
1
1
2
2
(15)
i
==
11
j
EP
=
r
c
[
]
∑∑
w
(
i
,
j
)
+
w
(
i
,
j
)
1
2
i
==
11
j
Q
(
i
,
j
)
Q
(
i
,
j
)
and
represently show the edge information retention of each
1
2
iw
reflect the absolute value of the gradient of each pixel.
Mean value, mean gradient, information entropy, deviation index and edge preserva-
tion all have corresponding improvements. The fused images' average energy of
wavelet coefficients is more than that of two original images. The experiment data are
shown in the Table 1. I1, and I2 represent two original images. TM represents the
traditional method. HI, LI and LHI respectively represent high-frequency coefficient
fusion improvement, low-frequency coefficient fusion improvement and
low-high-frequency coefficient fusion improvement.
w
(
i
,
j
)
(
,
j
)
pixel.
and
1
2
Table 1
. Comparison of fusion methods
IF
UF
CE
MV
MG
IE
DI
EP
WAE
I 1
98.3719
48.3449
7.4612
1.3037
0.0314
I 2
98.3657
49.7548
7.4352
1.3582
0.0291
TM
98.5281
53.2725
7.4462
1.2221
0.2140
0.0565
HI
98.5296
53.2876
7.4453
1.2216
0.2142
0.0565
LI
98.5517
65.3368
7.5029
0.0531
0.2300
0.0565
LHI
98.5520
65.3482
7.5019
0.0530
0.2301
0.0565
5 Conclusion
To sum up, the relevant experiments show that the image fusion algorithm based on
discrete wavelet transform and canny operator can be used on multi-focus images
fusion, which enables to achieve a well-effect fused image with higher clarity and
improve the regarding performance index. It makes good use of local time-frequency
characteristic and keeps image edge information better.
Search WWH ::
Custom Search