1
i
X
I
I
In DMP, the most significant and delicate stages are the
preparation, the mensuration and the subsequent trans
formations.
PREPARATION
Control data, correction data
Existing maps (or data files)
Photographs
FIELD: verification
completion
MENSURATION on:
Existing maps
sequence may
be reverse
Control data
(in map
photo
digitiser
system)
Changed data
files (in
photo
digitiser
system)
~T
DATA EDITING
CONDITIONING;
(for transformations)
Conditioned control data
Edited changed data files
(in both digitisers' systems)
TRANSFORMATIONS corrections) of:
control data to map (or field) system,
changed data to photo system
EXTERNAL ORIENTATION
i
Rotation matrix
TRANSFORMATION*corrections)
Photo to model
TRANSFORMATION corrections)
Model to map (or field)
t
Changed data files
PREPARE
Graphic changed data
files
PLOT
changes on existing
map
UPDATE:
Basic Data Files
PREPARE
new graphic data files
PLOT:
new map
FURTHER CARTOGRAPHIC
TREATMENT
Fig. 4. Sequential stages of updating by DMP.
Requirements
Before discussing the main stages of the updating pro
cess, it seems useful to review the requirements for digi
tal procedures. The following list is restricted, however,
to the operational characteristics and performance.
Compatibility with the present and future data base
technology, in particular the software and the data
input and output;
Comprehensive and efficient identification and ex
traction of changes;
Phased and parallel operation, and time-lineness in
data flows;
Applicability of the existing equipment;
Simple operation and support;
Quality of the existing data base should be main
tained or improved.
The sequence of these requirements is arbitrary.
Two realms in data flow
The data and operations can be virtually separated into
the semantic and the geometric parts. The semantic
realm covers operations such as identification, extrac
tion, encoding and indexing, and symbolising data
items, whereas the geometric realm concerns the locati-
onal data of the elements of semantic information, the
associated transformations and corrections, and posi
tioning. In the following, the operations of the two
realms will be separately reviewed.
Flow of semantic information
Changes
Though the identification and extraction of changes is
not unique for DMP, some attention is given to it in this
paper because it tends to dominate the overall perfor
mance and reliability of updating. The primary data
source are new survey photographs (or other images).
Before discussing the extraction of relevant changes,
however, it seems appropriate to review some related
issues concerning the changes in terrain. Most of the
factors listed in figure 5 can be differentiated further,
though this is not necessary in the context of the paper.
Changes
in terrain
Natural
phenomena
Human
activities
Errosion
-Desertification
-Climatic change
Disasters
Other
Physical
Administrative
Legal
Other
-Relevant
—Partly relevant
-Non relevant
Dynamic
view
_Gradual
changes
Sudden
changes
Fast
Medium
Slow
Very slow
- Disasters
- Polytical
- Legal
- Other
Predictable (deterministic processes)
-PredictabilityPartly predictable (stochastic processes)
Non predictable (e.g.disasters)
Fig. 5. Issues related to changes in terrain,
information flow
Figure 6 shows the flow of the semantic information.
The corresponding operations are involved in all stages
of the process and several of them (indexed by M) coin
cide with operations in the geometric data flow (figure
10; for S see also figure 10).
^Existing mapTI
files (or old
Ipho tographs )j
Other documentsi
New photographs
(or other lmages)j
Inputs
'Preparation
Encode mark new [S
features
.Editing and
generallsa-
Plotting
Prepare [~\j"
for drafting
Enter code
Interprete [m
Symbolise Items,
enter text numbers
Interprete changes
extract (relevant)
Generalise[M_
ïatlc Information)
Verify correct
(semantic Information)
Prepare, analyse Inputs
perceive changes
Fig. 6. Flow of semantic information.
The most delicate operations in the semantic realm are
change identification and extraction of the relevant data.
Considerable effort has been made to automate these
operations, though with little success so far.
296
NGT GEODESIA 82
