generated using different Ground Sampling 
Distance (GSD). 
Manual DTM GSD = 28 ft. 
Automatic DTM GSD = 25 ft. 
No. of DTM points Compared: 744 
Average Error Calculated: -0.08 ft. 
Standard Deviation: 1.68 ft. 
Manual DTM GSD = 25 ft. 
Automatic DTM GSD = 12.5 ft. 
then interpolated to 25ft. 
No. of DTM points Compared: 744 
Average Error Calculated: -0.035 ft. 
Standard Deviation: 1.28 ft. 
There are two conditions under which the above 
accuracies were obtained. First, ground controls 
were minimal in all above models, and second, 
images were scanned at a 22.5 micron resolution. 
With photo scale at 1"=300’, one pixel is 0.25 ft 
on ground. Scanning with 12.5 micron resolution 
would have yielded 0.15 ft on ground, a 
significant difference. 
3.4 Speed 
An average of twenty percent of all models seems 
to be covered by elevated vegetation. Manual 
editing or manual extraction would be needed to 
obtain correct elevations of ground surface. This 
task is performed within the manual 
extraction/editing option available in most softcopy 
photogrammetric systems. The time needed for 
the additional manual extraction conforms to 
normal production standards of manual DTM 
extraction, assuming conventional 
photogrammetric systems. This is considered a 
loss in the speed and efficiency gained by the 
automated DTM extraction capability of the 
softcopy system. 
Durations were estimated for operator-system 
interaction required for standard 
stereophotogrammetric procedures. An overall 
estimate was recorded for the entire end-to-end 
process in the Intergraph system. A detailed step- 
by-step breakdown periods were recorded for 
Helava. On both cases, interaction durations were 
superior to what is expected in a conventional 
photogrammetric machine. 
132 
3.2 Expertise 
There is a shift in the kind of expertise required to 
operate a softcopy photogrammetric workstation. 
This expertise is a departure from what is 
normally required for conventional 
photogrammetric systems. Characteristics of this 
shift can be summarized as follows: 
- For operators, less knowledge required in 
photogrammetric engineering, as a result of the 
"black box" concept, while more required in 
computers, Unix, windows, file systems, disk 
storage, stereo display, etc. 
- For operators, basic system training is sufficient. 
- For team leaders, high level training is required, 
to ensure smooth flow of operation when operators 
encounter a difficult, out-of-usual, situation with 
system software/hardware. 
4. CONCLUDING REMARKS 
-It has become obvious that  softcopy 
photogrammetry technology does meet 
performance standards, even for large scale 
mapping, such as WisDOT photogrammetric 
operations. 
-Substantial gain in speed can be achieved with 
softcopy photogrammetric systems, assuming other 
bottle necks, such as data volume, are handled 
efficiently. 
-High end systems do provide at least comparable 
accuracy, when automated extraction methods 
work. However, substantial enhancements are 
needed to add some intelligence in the extraction 
process. Matching techniques tend to fail with 
simplest unexpected conditions. 
-Conversion of conventional production facilities 
into softcopy operations requires significant re- 
training of operators. Cost issues, data volume, 
technical SW/HW support need also to be 
carefully considered. 
ACKNOWLEDGMENT 
This paper deals with aspects of a research project 
funded by the Wisconsin Department of 
Transportation, under project number 0092-35-57. 
Évalua 
co 
Mots-C 
Le dom: 
profonc 
sont In 
nombre 
caracté 
des ut 
Systeme 
les me 
entrepr 
des re 
équipen 
conséqi 
directi 
mais 
importa 
d'évali 
Systéme 
traditi 
indique 
Systeme 
photogr 
entrepr 
photogr 
La sél: 
précis, 
restrei 
volonté 
object: 
prélimi 
systeme 
aux sy: 
certain 
examiné 
Opérate