ling, skeleton
attributes refer
on and contour
cluding relative
can be derived
The first issue
y means of an
method using
aches allow to
S of statistical
more relevant
is context, the
representation
ized simulated
modelling was
| densification.
S but also real
, and the rules
|. The analysis
logic model.
elective with
portray terrain
indancy of the
expert system
rrain features,
is practically
rain surface in
ximateld by a
(via different
inusoids with
. Therefore
Yorphology by
itives, and the
the geometric
atical definition
issimilation to
ose are: semi-
face, conical
loidal surface,
ia geometric
signal to the
n morphology
representation. The probability of simulating perfectly
terrain morphologic features by the geometric primitives is
very low, but combining those primitives will increase the
probability. In any case, the verification of the concept by
real terrain morphologic features is necessary.
Basically, terrain morphologic modelling can be performed
in the following manners:
- Selective representation of terrain morphologic features.
- Semi-automated representation of terrain morphologic
features.
- Combination thereof (optimum representation).
1.1 Selective representation
This method is carried out manually to portray the terrain
morphology. It is applied to abrupt changes in terrain slope.
Basically, it is a subjective method of portraying the
skeleton of terrain morphology.
The main stages of optimum representation are shown in
figure 1.
SELECTIVE EXTRACTION AND
SEGMENTATION
SELECTIVE REPRESENTATION
EVALUATION
Fig. 1 Main stages of selective representation
The general procedure for data preparation and feature
extraction for selective representation of distinct
morphometric features (X information) is treated by
Makarovic (1976).
Because the procedur is subjective, it needs to be
systematised. To attain a balance between selective and
semi automated representation via a smooth operation,
some rules have been formulated. These represent the
RULE BASE for terrain morphologic representation.
The general procedure for segmentation, extraction and
selective representation of the terrain morphologic features
are explained by Charif (1991). From the results of the
experimental tests applied to ideal geometric generated
primitives, their composite surfaces and to real terrain
surfaces, some rules have been extracted for selective
representation of the terrain morphologic features (Charif
and Makarovic, 1992).
1.2 Semi-automatic representation.
This is a method for representing terrain regions, which are
mainly homogeneous, though irregular, thus providing the
filling information ( M information). The density of the grid
is locally adapted to terrain morphology.
The main stages of semi-automated representation are
shown in figure 2.
SEMI AUTOMATED REPRESENTATION
EVALUATION
Fig. 2 Main stages of semiautomated representation
In Makarovic (1973), a on-dimensional (1D) Laplacian
operator was used separately in the X and Y directions.
Tests using some representative, geometrically ideal
primitive surfaces show that "1D-Laplacian in four
directions" proves to be a potential alternative criterion for
the self-adaptive densification in semi-automated
representation Charif (1992)
For the study, the following densification criteria were used:
1D-Laplacian algorithm separately in X and Y, 2D-Laplacian
algorithm, extended 2D-Laplacian algorithm, and
1D-Laplacian algorithm separately in four directions.
The following potential alternative criteria: median height,
fitted plane, and second difference for a quadruple of
points, separately in the X and Y directions, should be
investigated, to define the optimum densification criterion
in semi-automatic representation of terrain morphologic
features.
1.3 Optimum terrain morphology representation.
This method concerns selective representation of distinct
morphologic features, followed by semi-automated
representation of more homogeneous morphologic terrain
features.
The four main stages of optimum representation are shown
in figure 3.
SELECTIVE EXTRACTION AND
SEGMENTATION
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
