sence of
terrain
1 points,
contour
trend of
aking of
1s later
nd the
yrdinates
n values
'ocedure
e object
points is
que was
id-white
dericton
taken at
e angle
spreads
River, is
ea level.
' incline
the city
canning
ScanJet
'ast and
er were
lize the
images.
vailable
tion was
) or 118
] size of
ground.
| to 256
exterior
rom l:1
al is 1.0
vation is
lifferent
Photo 1.
] streets,
a sports
ich were
then computed using the correlation and the
space intersection software, and compared
with the coordinates measured on the 1:1 200
scale maps [Chen, 1993].
3.2 Test Results and Discussion
The results of the test are summarized in
Table 1.
The RMSEs are the root mean square errors
of the planimetric coordinates obtained by
the pseudo-stereo mapping technique with
respect to those measured in the 1:1200 scale
maps. The notation "26 « 1.5 m", etc. means
the percentage of the check points with
differences. smaller than:.1.5. m, etc... The
photographs for this test were taken from a
block flown for 1:10 000 scale mapping.
Therefore the assessment of the results are
based on the degree of compliance with the
map accuracy standards applicable at that
scale.
The accuracy specification for the "Urban
and Resource Digital Map Base" in the "Land
and Water Information Standards Manual"
states [NBGIC, 1991]:
Ninety percent (90%) of all “well defined
features” must fall within the positional
accuracy ...[2.5m]. Well-defined features are
those whose positional accuracy is not
adversely affected by vegetative cover.
Accuracy of the digital data (point, line,
area) can be defined as the difference
between the position of the associated data in
the digital file and the real position of the
represented features on the earth.
Although this requirement is rather strict,
87% of the well defined feature points
satisfied it. This is only 3% below the 90%
tolerance. For graphical map production, the
generally accepted tolerance in North
America is 0.5 mm at publication scale, at the
9096 confidence level. This corresponds to
5.0 m for the 1:10 000 scale. All points were
Within this tolerance.
The following points should be considered to
put the results in proper perspective:
l. The pixel size of the digitization was 3.0
m on the ground, which is larger than the
2.5m tolerance set by the map accuracy
standards.
2. The stage plate of the scanner used was
8.5 x 14 inches, which is narrower than
141
the size of the photographs. Thus, two of
the primary corner fiducial marks and one
of the secondary marks along the edges of
the photographs were missed. This
introduced an uncertainty in the interior
orientation.
3. The particular scanner employed is only
suitable for digitizing opaque
photographic prints. The poor
dimensional stability of paper is of course
a source of significant distortion.
4. Because of storage limitations in the
computer, only the overlapping areas of
the photographs were kept on the files
and used for selecting GCPs. Therefore,
the geometry of the bundle of rays in the
space resection were not the most
favorable.
5: The'6l:15:200 scale maps: used for
comparison was accepted as error free.
Under the above circumstances the test
results can be considered satisfactory and
accepted as a proof of the feasibility of the
pseudo-stereo mapping technique developed.
Further tests are of course needed to fully
validate and improve this technique.
4. CONCLUSIONS
The pseudo-stereo mapping technique is
applicable when no orthoimages or DEMs
are available, or when the DEM is out of date
and the relief displacement is too significant
to be ignored. This technique has been
successfully implemented in a GIS as proven
by the outcome of an experiment. The ability
to perform soft copy mapping in a GIS has
the advantage that both the image and the
map reside and are manipulated in the same
workstation.
It is true that the results of the test do not
entirely satisfy the map accuracy standards
set for the digital base map production.
Firstly, the experiment was conducted under
less than ideal conditions for precision
mapping, as explained earlier. Secondly, the
objective of this development was to provide
tools for thematic mapping, performed by
non-photogrammetrists, when accuracy
requirements are not as stringent as in basic
mapping. Thirdly, simplicity and low-cost
were also objectives. This explains the