The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
1075
P'PS = /z-tan or, (5)
D = POP' + P'PS = (h/tana 2 ) + {h• tana x ) (io)
D = POC P' + P'PS = {h-tan a x ) + (h/ tana,)(6)
h _ D tan^) ^sin (2or,)
tan 2 («!) + l 2
(7)
3.4 Height estimation from combined SAR/optical data
As it will be shown in this paragraph, estimates for the heights
of elevated objects can also be deduced from combined optical
and SAR data. Simply, the distance between the displaced
bridge in the optical image and the displaced bridge in the SAR
image has to be measured.
OPT SAR
Figure 8: Height estimation of an elevated object by means of
combined optical and SAR imagery
In Figure 8 an elevated object P with height h is imaged by both
an optical sensor (OPT) and a SAR sensor (SAR). The two
sensors acquire the object with two different off nadir angles 0]
and 0). The corresponding depression angles are cii and a2.
Object P is projected to two different locations in the images
due to the different viewing geometries of the sensors. On the
one hand, the optical sensor measures angles and hence P is
imaged to point PO in the optical image. On the other hand, the
SAR sensor measures distances from a side-looking perspective.
Therefore, P is projected to point PS in the SAR image
(layover). Object P’s displacement can be expressed as (8) in
the optical image and as (9) in the SAR image (assuming a flat
wave front). The measured distance D between the displaced
object P in both images is the sum of the two separate
displacements in the optical and in the SAR image. Adding the
two separate shifts leads to equation (10). Finally, the elevation
of object P can be estimated with (11) (Inglada & Giros,2004).
(8)
(9)
h = D
tan (a 2 )
tan(a,)-tan(« 2 ) + l
(ii)
These equations have shown that only three parameters have to
be known in order to estimate the height of an elevated object
from combined optical and SAR imagery: the incidence angle
of each sensor and the distance between the displaced object in
the optical image and in the SAR image. In our case we assume
that all terrain in the imagery has already been accounted for by
ortho-rectifying the images with a DEM since we are only
interested in the bridge’s height. None-the-less, equation (11)
also allows for the estimation of three-dimensional surface
models in case no DEM is available. In such a case the images
would have to be registered using a local transformation. A
disparity map could be drawn and the terrain heights could be
deduced from the displayed shifts. Again, for such kind of
application a locally flat earth has to be assumed between the
projections PO and PS of object P.
4. SUMMARY AND FUTURE PERSPECTIVES
A comparison of bridges over water and bridges over land has
shown that the appearance of bridges in SAR imagery strongly
depends on their environment. Due to multi-path signal
propagation at bridges over water, a height value can be
determined directly from a single image. Such multiple bounces
do usually not occur at bridges over land. First concepts of this
long term project now have shown that a three-dimensional
modelling of elevated man-made objects is possible from a
single SAR image and from combined optical and SAR data.
In a next step the equations shown in this paper will be used in
order to obtain estimates for the railroad bridge near Zellingen
(Figure 4, 5, and 6). A field campaign will be done in order to
measure precisely the bridge height at each pillar. Hence,
estimates obtained from the imagery can be compared to the
real heights and thus be evaluated.
So far all displacement measurements are done manually.
Further steps of this project will be the integration of automatic
feature extraction and object based image registration
techniques. The bridges will be modelled on object level.
ACKOWLEDGEMENTS
The authors thank Dr. Essen, Dr. Schimpf and Mr. Brehm of
FGAN-FHR for providing the SAR test images of the
MEMPHIS sensor and Intermap Company for providing the
InSAR data set.
REFERENCES
Cadario, E., Schulz, K., Gross, H., Hammer, H., Thiele, A.,
Thoennessen, U., Weydahl, D.J., Soergel, U.,2008. Feature
extraction and change detection for bridges over water in
airborne and spacebome SAR image data. In: Proceedings of