4. CONCLUSION
Figure 3d. Fused image by the morphological pyramid
3.1 Visual examination
From the results of the image fusion presented in Figure 3c and
Figure 3d, Visual comparison suggests that the fused image
improves sharpness compared to the original images. Moreover,
compared with the result of the fusion obtained by the
Laplacian pyramid, the result of the fusion obtained by the
morphological pyramid is better in visual effect. Due to
limitations of human vision in terms of distinguishing the
number of grey levels, comparison and appreciation by visual
methods does not reveal the exact potentials of the fusion
methods. Hence, comparison of image statistics is attempted to
evaluate the results obtained by the Laplacian pyramid fusion
and the morphological pyramid fusion.
3.2 Quantitative analysis
In addition to visual examination, we conduct a quantitative
analysis. The analysis of the experimental results is based on
entropy and average gradient of an image (Table 1). The
entropy refers to the amount of the image information content.
The higher entropy, the better the image quality. The average
gradient refers to the sharpness of the image. The higher
average gradient, the better the image quality.
In Table 1, we show that entropy and average gradient of the
fused image by the morphological pyramid method are bigger
than these of the fused image by the Laplacian pyramid method.
From these results, we can conclude that the morphological
pyramid method provides more detailed spatial information
than the Laplacian pyramid-based fusion method.
Image
Entropy
Average gradient
Image focusing on the
left
7.09
3.46
Image focusing on the
right
7.11
5.06
Fused image by the
Laplacian pyramid
7.12
5.21
Fused image by the
morphological pyramid
7.13
5,23
Table 1. Evaluation of fused image by the Laplacian pyramid
fusion and the morphological pyramid fusion.
We have presented a region-based morphological pyramid
fusion method. With the experiments carried out on multi-focus
images, Compared experimental results pertaining to indicators
of entropy and average gradient, the fusion method based on the
morphological pyramid provides better visual and quantitative
results than that of the Laplacian pyramid-based fusion for
multi-focus image fusion. The implementation of our algorithm
is still in a preliminary stage. However, the results obtained so
far suggest that approach may be useful for the multifocus
image fusion. Investigate will be more thoroughly in the future.
REFERENCES
Blum R. S., and Zheng Liu, 2005. Multi-Sensor image fusion
and its applications. CRC Press in the special series on Signal
Processing and Communications, pp 1-29.
Burt P. J. and Andelson E. H., 1983.The Laplacian pyramid as a
compact image code. IEEE Trans, on Communications, vol. 31,
pp. 532-540.
Burt P.J., Hong T.H., and Rosenfield, 198 LA segmentation and
estimation on image region properties through the cooperatives
hierarchical computation. IEEE Transactions on SMC, vol. 12,
pp802-809.
Pham D.T., Aslantas V., 1999.Depth from defocusing using a
neural network. Pattern Recognition, vol. 32, pp. 715-727.
Piella G., 2002.A general framework for multiresolution image
fusion: from pixels to regions. Research report on CW1,
Amsterdam.
Toet A., 1989.A morphological pyramidal image decomposition.
Pattern Recognition Letters, vol. 9, pp. 255-261.
ACKNOWLEGEMENT
The authors would like to express their sincere gratitude to the
anonymous referees for pointing out several typos and for some
very helpful comments.
MONITO
U.S. Geological
KEY WORDS:
ABSTRACT:
Interferometric s;
deformation and
sensing Satellite
Japanese Earth R
Satellite 1 (Radar
seismic, and hydi
after eruptions pr<
and temporal pat
dynamics and est
hazards. Mapping
fundamental data
dynamic water-le’
flood events over
wild fire manager
natural hazards an
Interferometric s>
use of two or mor<
same area to extr:
patterns. A SAR
wavelength that c
centimeters and 1
under all weath
technique (Curiam
and phase of the
each ground resoli
can be calculated
that represents tl
amplitude or int
primarily by tern
constants, wherea;
primarily by the d
ground targets,
electromagnetic w
precise distances
resolution elemen
subtle change in el
InSAR is formed
from two spatially
separation of the 1
antennas may be n
interferometry, wf
and spacebome
(TOPSAR) and the
systems, which vs
high-precision dig
