PDATE
discussed.
presented.
in-house
topology.
re now to
y accurate
1damental
ctor data
erties and
it Vector
‘action as
facilitated
ient, by a
roduction
itionwide
ew of its
vorkflow,
trol.
cess for
ich was
database
m aerial
10dology
in a 2D
ns from
ROL
10000 is
designed
denmark
overs all
red and
denmark
through
escribed
| in the
^ "data-
s “data-
rocesses
' to the
eo
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
2.1 Data-collection workflow
The data-source for TOP10DK was B/W aerial photos in the
setup phase until 2000, and since then color photos - both in
1:25000. Photos are collected with 60% length overlap and
20% side overlap to allow a stereoscopic data-extraction
process. Photographing occurs from March through May
when Denmark is no longer covered by snow and the trees
are not in leaf. Data-collection covers approximately 1/5 of
Denmark every year. It is done by subcontractors after
detailed specifications (KMS, 2004a).
When weather and sun/light conditions are satisfactory, the
aerial photos are collected from a given flying height at
specified geographical coordinates given by KMS. Since year
2000, the aerial photos are scanned in 21-25 micron as
specified (KMS, 2004a), allowing data handling from then on
to be done digitally. Contractors are responsible for film
exposure, development and scanning and therefore for the
end-quality of data to be used for the object extraction.
Contact copies, dia-positives and enlargements were
produced up till 2004, but future data-extraction and control
will be pure digital.
2.2 Data-extraction workflow
Extracting data or objects from the photos, is done on
photogrammetric workstations by subcontractors. In the
database setup phase, contractors received film material and
other necessary data for the aerial triangulation (camera
calibration, GCP’s etc.). The established photogrammetric
parameters (tie-point coordinates, sketches etc.) were then
returned to KMS together with the extracted objects.
For the database update phase, subcontractors receive both
new digital photos (images) and the aerial triangulation
parameters originating from the first data production in that
area - to be able to set up the exact same aerial triangulation
for the new images.
Map-objects are extracted, in all tree dimensions, according
to rules specified in an object specification (KMS, 2004b). A
new updated digital version of the object specification is
scheduled for every years production — as changes in the
object rules are still in progress. Object-updates must take
into account, both changes derived from renewed
specification rules as well as physical changes in objects.
Physical changes, i.e. new, changed or erased objects, when
compared to the current database content. For this purpose,
contractors receive all existing objects in their area from the
database, allowing them to impose vector data to the images.
2.3 Data-control procedure
A number of control processes are implemented in the
production flow. They are developed in different stages of the
database setup and the database update phases - and for
different needs. Most of them are incorporated in Mapcheck,
which is an in-house developed control- / GIS- software used
by both KMS and the subcontractors. Other control processes
are executed in other surroundings or performed manually.
753
2.3.1 Control of photos — The data-collection worktlow
includes different control processes. These processes are to
ensure that data is adapted with sufficient good quality for the
map-object extraction. The control processes, which are all
performed by in-house employees, treat subjects as:
- Automatic mathematical control of photo-point accuracy
and overlap between neighboring photos (by exposure
GPS coordinates according to photo-points as specified)
- Visual control of developed film for visibility, color and
coverage (clouds or shadows) and photographic end-
products for visual appearance
- Random check of scanned images. Visually control for
color balance, hot spots etc. and automatically control
for color distribution (both since 2002).
2.3.2 Control of objects. The data-extraction workflow also
includes different control processes. These are adapted to
ensure that objects are collected according to specifications,
and keep the specified geometric accuracy. The control
processes are performed by in-house employees and treats the
following subjects.
- In the database and aerial triangulation setup phase,
independent control of the geometric accuracy was
performed by large-scale photos in 1:5000 in test sample
areas. In these photos selected objects were measured at
photogrammetric workstations and compared to the
extracted data from 1:25000 photos, for statistical
judgment.
- In the database setup and update phases control did
and still does concern topics such as:
- Topology in data, is evaluated by function that
automatically check if all vectors are connected as
specified
- Individual objects extracted according to
specifications. Are checked by functions that
automatically measure/calculate object “values” and
compare these to specification values
- Data completeness and geometry is checked by
manual control. Objects are compared visually with
their appearance in an image on a PC consol.
Control is carried out on polynomial rectified
images, with low correlation away from the photo-
center
Because all production is done by different subcontractors,
the control procedure is taken into evaluation for every
production block.
2.4 Evaluation of control procedure
During development and production of TOPIODK, all control
processes have been in focus and adjusted. Because
TOPIODK is not of a predefined design, it is an ongoing job
to develop both specifications and the control routines,
according to current production-workflow and the current
data quality.
2.4.1 Evaluating, control processes. From the current state
of data in the database and the existing control procedures
referred in 2.3, ithas been possible to identify the following
list of areas where suggestions for the control processes or
the workflow, would add value to the data quality.
