23,
} a
an
nice
cal
0,
93)
ve
nap
be
ail
lue
>re
che
of
ita
"th
to
^om
Lng
1ce
)n.
^io
al
to
‚er
Lse
1ts
ic
"ue
1S,
‚he
ice
er
he
The incidence angle of the radar beam component at the
ground element can be reconstructed, and .the angular
dependency on backscatter derived. Based on this
information, a SAR image can be radiometrically corrected
with respect to terrain relief influences on the pixel grey
values.
Layover and shadow areas, which only contain contorted
or indirect information, can be numerically determined and
coded, so that they can be excluded from or weighted in
evaluations relying on statistical grey value information.
2.2 Image-to-Map Registration
The aim of this geometric approach is to generate a
radar image which is geometrically independent on the SAR
imaging process (removal of geometric distortions), and in
doing so, transform the image into the geometry of a
selected map projection. Moreover, the correct positioning
parameters of the image within a global map coordinate grid
have to be identified.
The final product of this process can be referred to as
a geocoded image, or, if the reference map shows an
orthogonal projection, as an ortho-image.
In fact, geocoding is performed by a complex multi-step
process which has been addressed in a paper by Domik et al.
1984, and is still a subject of comprehensive investigations
(Raggam et al. 1986). For the ease of reference a brief
description of two alternative principles for geocoding is
given in the following.
Geocoding Based on Simulation
Simulation techniques can be used to transform small
object space segments ("chips") from a data base containing
selected topographic and thematic features into a
specifiable SAR image geometry. Criterion for the selection
of these chips is a good and possibly automatic geometric
comparability of the simulated chip and the homologue
feature in the SAR image. Residual location differences
between simulated and real feature measured in the image
space are used to determine the coefficients of a polynomial
relative rectification of the real image. In this
procedure, the real image is transformed into exactly that
SAR image geometry which is consistent with the
radargrammetric model. Eventually, the image-to-object
space transformation and subsequent resampling yields the
geocoded image product.
325