CMRT09: Object Extraction for 3D City Models, Road Databases and Traffic Monitoring - Concepts, Algorithms, and Evaluation
156
inherent noise is reduced and underlying structures are enhanced
depending on their length, their orientation or their intensity.
In the image enhancement context this approach is most suitable
for fine-structured areas, e.g. city centers. The main problem
lies in the determination of thresholds for suppression and em
phasis of structures. The determination of the threshold and the
number of coefficients respectively is still experiential and highly
dependent on the image content. If the scenes are reconstructed
by a fix number of coefficients, the complexity of the scene is
restricted. As the image description by the curvelet coefficients
is purely based on structures, by omitting coefficients originally
smooth areas are often affected by artifacts. At the moment the
quadratic weighting of the single curvelet coefficients seems to
be the best solution for fully automatic processing chains.
The change detection approach provides excellent results in ur
ban areas. The great advantage over pixel based methods is the
sensitivity towards changes in structures and the possibility to
predefine the scale and the strength of changes to be mapped.
Problems occur in natural surroundings like forested areas, where
the status of the foliage has an important seasonal impact on the
backscattering behavior. Not to mention the weather conditions,
snow cover with different moistures can highly modify the ap
pearance in a SAR image. In consequence of that the interpre
tation of the detected changes is very challenging. Although the
change images contain clear structures without any disturbances,
it is nearly impossible to distinguish man-made from natural, e.g.
seasonal, changes, without a priori knowledge about the land
cover.
As the present results proved that two single polarized SAR im
ages can be used to indicate changes happened to the imaged
area, but they do not provide the information needed to interpret
these changes, our future research will try to include other data
sources into the processing chain. To discriminate natural cover
from man-made objects, a coherence layer, that exploits the phase
information of the input images could be helpful. Polarimetrie
layers could facilitate the interpretation by attaching information
about the scattering types to the detected changes. Apart from
remote sensing data it is quite conceivable to introduce a priori
knowledge by overlaying the change layer with land cover classi
fications from optical data sources as well as with cadastral data
sets.
References
Ali, S. M., Javed, M. Y. and Khattak, N. S., 2007. Wavelet-
Based Despeckling of Synthetic Aperture Radar Images Using
Adaptive and Mean Filters. In: Proceedings of World Academy
of Science, Engineering and Technology, Venice (Italy), Voi. 25,
pp. 39-43.
Balz, T„ 2004. SAR simulation based change detection with
high-resolution SAR images in urban environments. In: IS-
PRS Congress, Istanbul 2004, Proceedings of Commission VII,
Voi. 35, pp. 472^77.
Bouyahia, Z., Benyoussef, L. and Derrode, S., 2008. Change
detection in synthetic aperture radar images with a sliding hid
den Markov chain model. Journal of Applied Remote Sensing
(JARS), SPIE.
Candès, E. J. and Demanet, L., 2002b. Curvelets and Fourier
integral operators. Compie Rendus de l’Academie des Sciences
336, pp. 395-398.
Candès, E. J. and Donoho, D. L., 1999. Curve and Sur
face Fitting. Innovations in Applied Mathematics, Vanderbilt
University Press, Nashville (TN), Saint-Malo (France), chapter
Curvelets - a surprisingly effective nonadaptive representation
for objects with edges, pp. 105-120.
Candes, E. J. and Donoho, D. L., 2002a. New Tight Frames of
Curvelets and Optimal Representations of Objects with Smooth
Singularities. Comm. Pure Appl. Math. 57, pp. 219-266.
Candes, E. J. and Donoho, D. L., 2003a. Continuous Curvelet
Transform II: Discretization and Frames. Appl. Comput. Har
mon. Anal. 19, pp. 162-197.
Candes, E. J. and Donoho, D. L., 2003b. Continuous Curvelet
Transform I: Resolution of the Wavefront Set. Appl. Comput.
Harmon. Anal. 19, pp. 198-222.
Candes, E. J. and Guo, F., 2002. New Multiscale Trans
forms, Minimum Total Variation Synthesis: Applications to
Edge-Preserving Image Reconstruction. Signal Processing 82,
pp. 1519-1543.
Candes, E. J., Demanet, L., Donoho, D. L. and Ying, L., 2005.
Fast Discrete Curvelet Transforms. Multiscale Model. Simul. 5,
pp. 861-899.
Demanet, L., 2007. curvelet.org. http://www.curvelet.org. (ac
cessed on 26 March 2009).
Derrode, S., Mercier, G. and Pieczynski, W., 2003. Unsuper
vised Change Detection in SAR Images Using a Multicompo
nent HMC model. In: P. Smits and L. Bruzzone (eds), Sec
ond International Workshop on the Analysis of Multitemporal
Remote Sensing Images, European Commission Joint Research
Centre, Ispra (Italy), pp. 16-18.
Donoho, D. L. and Duncan, M. R., 2000. Digital Curvelet
Transform: Strategy, Implementation and Experiments. In:
Aerosense 2000, Wavelet Applications VII, Orlando (FL),
pp. 12-29.
Gleich, D., Kseneman, M. and Datcu, M., 2008. Despeckling
of TerraSAR-X data using second generation wavelets. In: ESA
EUSC 2008: Image Information Mining, Frascati (Italy).
Ma, L., Ma, J. and Shen, Y., 2007. Pixel Fusion Based Curvelets
and Wavelets Denoise Algorithm. Engineering Letters, On
line Journal of International Association of Engineers (IAENG)
14(2), pp. 130-134.
Marcos, J.-S., Romero, R., Carrasco, D., Moreno, V., Valero,
J. L. and Lafitte, M„ 2006. Implementation of new SAR change
detection methods: superresolution SAR change detector. In:
ESA-EUSC 2006: Image Information Mining for Security and
Intelligence, Torrejon air base - Madrid (Spain).
Moghaddam, H. A., Zouj, M. J. V. and Dehghani, M., 2004.
Bayesian-based Despeckling in Wavelet Domain Using ”a
Trous” Algorithm. In: ISPRS Congress, Istanbul 2004, Pro
ceedings of Commission VII, pp. 27-13.
Polidori, L., Caillault, S. and Canaud, J.-L., 1995. Change de
tection in radar images: methods and operational constraints. In:
Geoscience and Remote Sensing Symposium, 1995. IGARSS
'95. ’Quantitative Remote Sensing for Science and Applica
tions’, Florence (Italy), Vol. 2, pp. 1529-1531.
Romero, R., M. J.-S., Carrasco, D., Moreno, V., Valero, J. L.
and Lafitte, M., 2006. SAR Superresolution Change Detection
for Security Applications. In: ESA-EUSC: Image Information
Mining for Security and Intelligence 2006, EUSC, Torrejon air
base - Madrid (Spain).
Sveinsson, J. and Benediktsson, J., 2007. Combined wavelet
and curvelet denoising of SAR images using TV segmenta
tion. In: Geoscience and Remote Sensing Symposium, IGARSS
2007, Barcelona (Spain), pp. 503-506.
Touzi, R., Lopes, A. and Bousquet, P., 1988. A statistical and
geometrical edge detector for sar images. IEEE Transactions on
Geoscience and Remote Sensing 26(6), pp. 764-773.
Wright, P, Macklin, T., Willis, C. and Rye, T., 2005. Coherent
Change Detection with SAR. In: European Radar Conference,
EURAD 2005, Paris (France), pp. 17-20.
Ying, L., Demanet, L. and Candes, E. J., 2005. 3D Discrete
Curvelet Transform. In: Proc. Wavelets XI conf. 2005, Sympo
sium on Optical Sciende and Technology, San Diego (CA).
