Environmental Engineering Reference
In-Depth Information
Footprint
extractor
footprint
Optical Image
Seismic
vulnerability
evalution
Principal
Directions
(by-product)
Vulnerability
estimate
# floors
Floor maker
detection
Rotation
SAR Image
FIGURE 5.6
An EO-based scheme for geographically extensive seismic vulnerability estimation.
work is to build a processing chain like the one outlined in
Fig. 5.6, where a fusion process is shown: a VHR optical and a
VHR SAR images are acquired over the same area and registered.
On every single building, a footprint extraction is performed
on the optical image; the building orientation is passed on to
the SAR analysis algorithm and used to seek lines of scatterers
indicative of floors. The floor count is then passed, together
with more relevant information extracted from EO data, to the
seismic vulnerability model. Ideally, this method should enable
production of seismic vulnerability maps useful for risk scenario
analysis.
This researchhighlighted various difficulties inherent in fusing
optical and radar VHR images. Among the most important
findings were:
andmethods exploitingmeter and submeter resolution radar data
had previously been developed in the airborne domain and could
in principle have been reused on spaceborne domain, a large gap
still remains probably due to the following reasons. First, despite
the significant progress in SAR systems development, the sheer
difference of two orders of magnitude in the instrument-target
distance unavoidably introduces a certain degree of dissimilar-
ity between the two types of data, and negatively biases the
quality of spaceborne data. On the other hand, pointwise acqui-
sitions of airborne data did not push towards the development
of suitable algorithms capable of exploiting long time sequences,
which become available with the new generation of spaceborne
sensors. Nevertheless, research is in progress and some results
are slowly coming up. The advent of satellite constellations like
COSMO/SkyMed that are capable of delivering large amounts
of data in a very short time will further thrust applications in
the domains of security and disaster management, especially in
terms of preparedness and vulnerability assessment, which is at
least as important as damage assessment in a global perspective.
In optical and radar images the effect of elevation generate
foreshortening effects that are opposite in nature; elevated
objects bend towards the sensor in radar images, whereas they
bend away in optical images. Since most objects in urban areas
are elevated, this phenomenon results in extensive mismatch
between object images.
●
Geocoding and orthorectification of images is frequently not
sufficient to match objects even in the absence of foreshorten-
ing effects (e.g., a horizontal shape at zero elevation). Different
acquisition geometries, different resolution of the images, dif-
ferent resolutions of the DEMs used, residual errors, all sum
up to a displacement between images of the same object in
the two types of data. The displacement may reach as high as
several pixels and needs to be corrected manually
a posteriori
.
●
Acknowledgments
The authors wish to thank Massimiliano Aldrighi for the data
processing resulting in images displayed in Fig. 5.2 and Fig. 5.3.
They also wish to acknowledge the support of the Italian Space
Agency and the Italian Civil Protection Department in providing
COSMO/SkyMed data and ancillary information.
Conclusions
References
In this chapter it has been shown that EO applied to urban areas
benefited in many ways from the recent refinement in spatial res-
olution of spaceborne radar sensors. Although some algorithms
Aoki, H., Matsuoka, M. and Yamazaki, F. (1998) Characteristics
of satellite SAR images in the damaged areas due to the
Search WWH ::
Custom Search