1.5 
| —e— fixed parameter filter | A 
= tT | t adaptive filter M Hie 
xr ——- 
T Sl eT 
Pa 
wu o5 |} 
0 
0.25 0.5 0.75 1 1.25 1.5 
Filtering parameter y 
  
  
  
Fig.13 Error characteristics of the adaptive filter 
and the fixed parameter filter(S/N=40dB). 
  
  
  
30 L i 
{ ' 
25 | — dà without filter . 
| » =X= wfixed parameter filter à 
20 | | —o— adaptive filter | 0 
  
Error(%) 
  
  
  
  
70 60 50 40 30 20 
S/N(dB) 
  
  
  
Fig.14 Error characteristics against S/N. 
  
  
  
  
  
  
  
  
  
  
50 — ew 
  
  
  
  
| 
| 
| 
| 
| 
il 9 
1 Piel 256 
CC — 
Fig.15 The results of 3-D analysis. 
f e Steep model 
Normal model 
ice Rlat model 
  
  
  
Error(%) 
Q = ND Wa Mo Jobo 
  
Fig.16 The error against 
the imaging angle parameter, B/H. 
  
remeron 
| —Steep model | 
| e=Normal model 
[| Flat Model - 
  
  
  
  
  
  
1 Pixel ^us 280 
Fig.17 The central cross section 
of terrain models. 
4.CONCLUSION 
The correlation coefficient analysis method is 
susceptible to noise in the flat land, conversely it 
is insensible with noise in the steep region like a 
top of mountain. Accordingly, we can obtain more 
accurate height data using adaptive filter which 
automatically select some different characteristic 
filters fitting to geographical features. Although 
the proposed method is the simple algorithm, it is 
effective to get accurate data. 
In the correlation analysis method, it is very 
important to decide to the imaging angle of 
satellite system. We can consider that more 
accuracy terrain data is restored under the most 
suitable imaging angle and the best size of the 
correlation window. 
Consequently, by using the proposed method, we 
can improve the performance of the system 
without expanding the optical aperture. 
REFERENCE 
[1]M. Lehner, 1995. The photogrammetric Eval- 
uation of MOMS-02 mode 3 data (Mexico, 
Ethiopia). Proceeding of SPIE Conference on 
Remote Sensing and Reconstruction for 3-D 
Objects and Scene, pp.102-114. 
[2]R. Nagura, 1992. High resolution stereo 
imaging and data compression.Proceedings of 
the Central Symposium of the 'International 
Space Year' Conference, pp.907-912. 
[3]R. Nagura, 1991. Conceptual study on 
forthcoming very high resolution visible and 
near infrared radiometers. Asian Conference on 
Remote Sensing, pp.E-3-1_E-3-6. 
314 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 
  
  
  
KEY 
ive me 
ABS” 
À new 
ution ( 
years. 
in the 
and la 
parcel: 
compl: 
impro 
feature 
unwan 
cover 
tential 
lysis o 
A str 
been c 
remov 
of an 
tion m 
more c 
lations 
urban 
At pre 
image 
provid 
atic M 
where 
tial ar 
been f 
and co 
use, ( 
applice 
tion of 
(Blami 
This h 
spatial 
that th 
them a 
to ber 
In the 
classify 
likely i 
tical s: 
data-se 
matic 1 
to bel 
new in