-L
-CI
-Er
/-
1
r
i
V!
X
Automatic conditioning of the input data for interactive
(man-machine) operation has been more successful.
The relevant data items, their mutual relations and codes
(and symbols for graphical representation) should be
specified for the original mapping. Before updating,
however, some of the specifications may be modified
and/or added.
Change perception-general
For visual change perception, different data sources,
physical means, and techniques can be used. Figures 7a
and 7b show multiple possibilities, though the automatic
feature extraction technique is not included as it is not
operational.
Photographs
Existing maps
or digital file
-contact copies
-scaled photographs
-orthophotographs
Geometry -
Single
Stereo
Unchanged
Scaled to photo (approximately)
Transfomed to photo (exactly)
Selected
-All
Fig. 7a. Data sources for change extraction.
Means and
techniques
- Hard copies
Optical
projection
Electronic
display
Mixed
versions
Separate lay out
Superposition (overlay)
Separate projections
Superposition
Flickering
Separate displays
Superposition
Flickering
EHard copy optical display
Hard copy electronic display
Optical electronic displays
Fig. 7b. Means and techniques for change identification.
It is beyond the scope of this paper to evaluate the diffe
rent variants of change perception. The „separate lay
out" and the „direct-" or „optical superposition" of the
new and old data sources, however, are the most com
monly applied. Flere, data preprocessing and display for
visual perception are essential.
The procedure for change perception and extraction of
new data includes the following stages:
Isolation of the areas of potential changes by
reconnaissance;
Preprocessing of the image data for these areas, e.g.
stretching contrast, filtering, geometric transforma
tions (for superposition), etc.;
Comparative analysis of the „new" and „old" data
sources and perception of the differences (i.e., of the
raw changed-data);
Interpretation of the differences and extraction of the
relevant changes;
Marking, encoding and indexing the changed data;
Verifying completeness and correctness of the
changed data.
The most pertinent operations are the comparative ana
lysis and interpretation which should imply a good
strategy. The latter, however, should reflect the specific
structure of the data base concerned. Though it is bey
ond the scope of this paper to discuss the interrelation
between the strategy and the data structure, it seems
appropriate to review some pertinent issues concerning
the data structures.
Data structures
The following are some key structures some of which
are interrelated:
a. Hierarchy of information items and of their compo
nents;
b. Relations among items and topological relations;
c. Data formats i.e., line, polygon (areas), and grid.
Figure 8 indicates a hierarchy of data entities and some
related properties. The entities can be further
subdivided.
entities
Areas
(polygons)
Networks
(lines)
- Points
Natural
Min-made
Natural
Man-made
Natura1
Man-made
-physical
-abstract
Fig. 8. Hierarchy of data entities and their properties.
Other hierarchial structures may be based on entries
such as importance or changeability of the information
items (and of their elements). In such structures, the
items can be given priorities, ranked according to the po
tential uses of the data base. Hierarchies can also be for
med of the areaentities and of the networks. Typical
examples of areal entities are the matrix stratification
(fig. 9a), contiguous irregular zones (fig. 9b) and un
related irregular zones (fig. 9c).
f
I
F---
J__
I
I
I
I
I
"I
i
I
I
i
i
I
"l
I
i
I
i
I
I -
y
Fig. 9a.
Matrix stratification.
Fig. 9b.
Contiguous irregular
zones.
Fig. 9c.
Unrelated irregular zones.
Relational structures and data formats should, in princi
ple, be defined before the original mapping. Neverthe
less, some new relations and format conversion routines
may also be added in later stages, which may contribute
to a gradual transition from a graphical to a digital data
base.
Techniques for change perception
For comparative analysis, the „new" and the „old" data
sources are displayed or laid-out (fig. 7b). If the two data
sources are superimposed and matched globally, the
geometries (inclusive scales) should be equal. If only
small zones are matched locally, an approximate equality
of scales suffices. In the case where the two data
sources are laid-out separately, the scales may differ.
In the first case, i.e., when the „new" and the „old"
data sources are globally matched, the geometry of one
data source has to be transformed into the geometry of
the other. Hence, either the new image must be trans
formed into the geometry of the map (or digital files) or
the existing map (or digital files) must be transformed in
to the geometry of the new image [5] [7].
Both cases require a substantial additional effort. If the
two data sources are superimposed, the image detail
may be obscured by the map symbols, text, hatches,
shades, distortions caused by generalisation, etc. [2],
Hence, a selective display of the map data is desirable.
NGT GE0DESIA 82
297
