com- 
the 
ele- 
The 
are 
sely 
As- 
de- 
r a 
area 
| of 
'ided 
‘trol 
s at 
70s. 
were 
ges. 
' the 
the 
sting 
  
  
  
  
  
  
  
Figure 6. Distribution of Control Points 
and Ground Scenes of Two Images 
40 points compared with 1:10,000 maps: 
. mX=8.13, mY=8.16, mh=5.71 
40 points compared with 1:50,000 maps: 
mX-11.90, mY-12.65, mhz6.92 
A line map at the scale of 1:100,000 was compiled 
with all index and partial standard contours, some 
habitations, roads, and hydrography features, etc. 
(Figure 7). In addition, a DEM was registered from 
the stereo images, and then used to generate the 
control data of ZS-1. An orthophoto (Figure 8) at 
the scale of 1:50,000 was printed, and the residual 
RMSEs (mm) according to 26 check points are 
mx=0.24, my=0.26 
The experimentation shows that the approach de- 
scribed in this paper is successful in processing 
SPOT imagery and the achieved accuracies might meet 
specifications even up to a scale of 1:25,000. Un- 
fortunately, the (ground 10m) resolution of SPOT 
images supplied currently is not clear enough to 
well identify many details, such as habitations, 
riverets, and village roads, which have to be pre- 
cisely located for maps of these scales, and conse- 
quently, using SPOT imagery for line mapping is 
partly restricted unless images of higher resolu- 
tion are available. Nevertheless, DEM registration 
and orthophoto printing are much effective. 
  
= 
7 
  
  
  
  
Figure 7. Partial Original Map (1:100,000) Compiled with SPOT Images 
283 
As 
ES 
tr 
en. 
niet 
a 
i 
i 
; 
: 
ÿ 
E 
e