owing the 
need of 
lucing the 
he lowest 
mage and 
found, its 
' pyramid 
ear in the 
edited on 
ct wrong 
as of low 
wards, a 
reference 
oints can 
>d: cross 
rrelation 
one pixel. 
s without 
on, least 
accuracy 
and have 
he image 
> known 
pace they 
e used to 
triangles 
ints to a 
: liminate 
d on the 
nalytical 
matches 
her noise 
ce points 
0 covers 
ssult, we 
| layer of 
| in any 
original 
ations to 
ack into 
je. This 
location 
mage is 
there is 
aid with 
| typical 
ot relief 
| DEMs 
digital 
an enter 
ropriate 
ita from 
ve is to 
olution, 
graphs. 
be very 
satellite 
ie DEM 
  
would still have a vertical accuracy of 10 - 15 meters, which 
hardly effects the planimetric accuracy of the final 
orthophotos. Thus, the user can create the raster DEM using 
SPOT imagery and then apply it to aerial photos for relief 
correction. The orientation parameters must be derived from 
both sensors. 
GENERAL PROBLEMS OF IMPLEMENTATION 
Conclusions 
The integration of digital photogrammetry and raster GIS is a 
very important enhancement of the way in which information 
is acquired for geographic information systems. Currently, 
the techniques to automatically derive digital elevation models 
and digital orthophotos in raster format are operational in the 
ERDAS GIS package. Many enhancements of this system can 
be envisioned. For example, it would be useful to use a real, 
3-dimensional monitor to view image pairs as stereo-models, 
in which the operator could trace lines in space with a 3- 
dimensional cursor. The automatic interpretation of raster data 
is of critical importance for combining raster and vector 
GIS's. Therefore, feature extraction algorithms are being 
developed which will permit the user to- automatically find 
lines in the images, and even interpret them as objects on the 
ground such as, roads, rivers or buildings. 
Another enhancement to speed up digital mapping is automatic 
line following in the digital images. For this procedure, the 
operator would select the starting point of a linear feature, 
such as a road or a river, and the function would automatically 
trace this line as long as it is unambiguous. 
In the future there will be many new developments in 
automatic data acquisition for geographic information systems. 
Software vendors have recognized the importance of 
integrating photogrammetric functions in a GIS and are 
working on modules similar to the one described here. 
Ultimately, this information extraction technology will also 
enhance the importance of satellite sensors and digital aerial 
cameras for small-scale mapping and geographic data bases, 
as well as for civil engineering applications. 
  
  
  
Figure 3 
897