£2 ; hr. va à : 3 x.
Fig. 6. ]. Synthetic line image generated by P. 6.2. es -equalized a A of Fig. 63. Histogram-equalized E ONERE of
Gibbs sampling from the prior distribu- TOPSAR scene of 94/08/05. TOPSAR scene of 94/08/05.
tion. Different grey values/colors show
different line states.
Fig. 6. " Son of he E SE -ratio line Fig. 6.3. Masimum- likelihood classification " Fig. 6.6. Result of M AP estimation using
detector. the intensity (without utilization of simulated annealing based on intensity
prior knowledge). and generic prior knowledge.
Fig. 6.7. Result of MAP estimation using | Fig. 6.8. Corridor along road center line given Fig. ^ 9. Result of MAP estimation using
simulated annealing based on inten- in a GIS. simulated annealing based on inten-
sity, coherence and generic prior sity, coherence, GIS information and
knowledge. generic prior knowledge.
320
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
KEY M
ABSTF
Autom:
project
Science
to dete:
from hi,
one of t
we pres
The rec
in 3-D
sue of |
mation
and tel
ual 3-C
requires
sive ins
sify anc
GIS is |
that gc
compor
interfac
Methoc
agery h
Digital
worksh«
of the
jective
extract
regions
vides c
tinguisl
smooth
made c
ate. He
ideas fc
In secti
the che
with di
vide a
of buil
methoc
other I
metric
reconst
cludes
eling ir