operation took the longest time; it had be be done
properly (to consider few points without disturbing the
graphical detail reproduction quality).
After all arches have been digitized, the corrections
operation follows (in fact, continuous lines polygons
should be closed). Topology should be also made.
As a matter of fact, during this stage, we can really point
out digitising errors using tolerances tested already
(Nipu C. & Nipu C.D,1992).
Another procedure element, which we should take into
account, during digitising operations, is the separation of
the themes (contained on the map) to be digitised as
much as possible. Digitising operations should be
ordinarily accomplished on the map manuscript; in this
case, some digitising transparencies should not be done.
Digitising by scanning followed by vectoring is a more
efficient procedure. Unfortunately, map manuscripts are
not always at our disposal. So, every important theme
should be copied on a transparency and digitised on
various layers later on. In this way, a topology for each
theme is to be made separately thus simplifying the
subsequent information analysing and plotting
procedures. As regards themes, such as, hydrographic or
road networks, a polygon-like topology should not be
established but a line-like topology is enough. Such a
well established topologic concept will make attribute
input in an easy way (elements featuring a certain
geometric shape: point, line, polygon).
3.4 Descriptive/Statistic Information Collection
Two methods were used to register attributes according
to the established nomenclature: (1) the geometric
element interactive selection using ARCEDIT function
by filling in attribute column; (2) INFO - type auxiliary
files containing a common column of a ARC/INFO
attribute file. So, the attributes could be separately on
files and automatically added to the corespondent object.
Both line-like attributes (on maps with a topological
grid, i.e. hydrographic grid of channels and road and
dike grid) and polygon - like attributes (on maps
showing land cover and hydrographic grid of lakes) have
been input, as well. The data bank will be subsequently
supplemented/completed with the statistical attributes
derived from some other sources.
3.5 Remote Sensing Data Integration: SPOT Image Case
During the first stage, we have tried to integrate data
derived from a SPOT image taken over in August 1990.
Owing to the huge informatic amount and the small
storage - computer capacity, we have chosen a subarea,
of the test field. Later on, this data base was completed
with other information derived from a 1992 SPOT
image. The images have been corrected and brought at
2B level by IGN Espace in France, within another
project. So, they have undergone the corrections: (1)
radiometric corrections to equalise/level detectors on
each spectral band; (2) geometric corrections due to
systematic distortions (the Earth rotation, panoramic
effect); (3) bi-dimension geometric corrections using
some ground control points.
During the first stage of Remote Sensing integration (see
Figure 2) we have selected a image window covering the
test area. During integration process, images could be
employed for two purposes: (a) for analysing needs when
image information should be correlated to the other data;
(b) for presenting needs when images are used as image
background nevertheless being a support in result
interpretations.
It is compulsory to do a vector - like integration when
analysing requirements are envisaged.
Employing only one multispectral image, we can get
more information plans, according to the investigated
topic. All these plans, i.e. raster - like classified images,
are made homogeneous and vectorial, being changed in
information plans suitable to GIS structure. During the
first stage, the hydrographic constituent was dwelt upon
also taking into account the great dynamic changes
arising from that information plan of the ever
changeable area (Denegre, 1991).
806
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
is
lil