sr to
posed
ilable
the
g the
jd by
ap is
ligital
f 25
. The
256
Leica
etric
age
ters
ints
The
gged
ious
photography. The DEM generated on a 50m grid
which resulted in 332 800 elevation posts with an
estimated RMS of 1.5m. Contours were generated
as well, and 3D contour inspection was performed
by draping the contours on the stereo-model for
stereo-interactive contour editing.
The orthoimage was produced pixel by pixel using
the collinearity equations and the gray values are
assigned using the nearest neighbour interpolation
resampling technique. The output ground pixel size
was set to be at 2m, which was determined as the
average value between the scanning and the human
eye resolution of 10lp/mm (Armenakis et al.,
1995).
To cover the entire map data set in a seamless
way, the individual orthophotos were mosaicked
and radiometrically corrected. Due to the large
size of the mosaicked raster file (208Mb) the
sheet was divided into four orthoquads with each
quad being a more manageable 52Mb. The
orthoquads were exported along with geolocation
data to the CARIS GIS system where they were
used as raster backdrops for the revision process.
4.2 Data revision
The revision of the 1:50 000 NTDB (National
Topographic DataBase) vector data based on the
digital orthophotomosaics was done using the
CARIS GIS. The metric accuracy of the registered
orthophotomosaics was evaluated by measuring
the coordinates of check points from the
orthophotomosaic and comparing them with given
values. For the four quads used for the revision of
the Jasper data, the standard deviation of the
coordinate differences were from 41.17 to
+1.67m in x and from +1.50 to +2.69m in y,
sufficient to meet the NATO A rating planimetric
accuracy requirement for the 1:50 000 maps.
The criteria for collecting new data or revising
existing features were based on the following
factors:
- the amount and type of change detected
- the accuracy of the existing feature
- the topology
- the feature morphology, and
- the significance of the feature.
The orthophotomosaic quads were displayed one at
a time and the CARIS file manager utility was used
to integrate the raster and vector data to
facilitate the use of superimposition for the
collection of the new data. A tile approach was
used for the revision process, where the operator
97
steps through the data set in small virtual map
tiles, revising all the features before moving to
the next tile. The task of visual change detection
was aided by relying on photo-prints from the
field verification for classification of roads, noted
additions, deletions and changes to features. The
zoom in/out capabilities were applied for data
collection. Image interpretation was
improved/assisted by employing interactive
contrast enhancement using either histogram
equalization or user specified histogram. During
revision the selective display of features was
applied for best results. In conjunction with the
display on screen, the operator used a stereoscope
and the aerial photographs with field information
to view the area's relief, identify features such as
watercourse, permanent snow and ice (glaciers)
and collect them with greater ease. Figure 2
shows the “old” and the revised “new” vector
data of the Athabasca River features.
The data set for Jasper map sheet was vectorized
cartographic data and required a slightly different
approach for revision than a positional data set.
When revising a positional data set, the exact
positional data would be collected. In the case of
the revision of a cartographic data set, features
that were not positionally correct due to
cartographic displacement are not re-positioned.
Repositioning these features would only result in
additional work at the cartographic editing stage,
although it may be considered to improve the
accuracy of the NTDB vectorized “old” data. To
determine if a feature was positionally incorrect
or just cartographically displaced, the
cartographic utility WYSIWYG was found very
useful. The WYSIWYG capability allowed the
operator to turn feature symbology on and view
the feature with its cartographic representation.
For example, a railroad that was positionally
displaced, but shown to be cartographically
correct with it's symbology displayed, was not
edited.
Attribute changes in the data were applied based
on information collected during the field work. The
height information of the planimetric features can
be derived from the DEM used to produce the
orthoimages either in real-time or in post-
processing mode. For the revision of this map, new
contours were automatically generated from the
DEM for evaluation purposes only, and the existing
metric contours were maintained.
The revised data was time-stamped. This enabled
the revision operator to identify all the features
that have been revised and separate them from the
original data for quality control.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996