International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
2. SEMANTIC MODEL FOR THE INTEGRATION OF
GIS-DATA AND AERIAL IMAGERY
Describing the integration of GIS-data with aerial imagery in
one semantic model is the starting point for object extraction, as
highlighted in detail concerning the extraction of field
boundaries and wind erosion obstacles in (Butenuth and Heipke
2004). The semantic model is differentiated in an object layer,
consisting of the real world, a GIS-layer, a geometric and
material part, as well as an image layer (cf. Figure 1). The
model is based on the assumption, that the used CIR-images are
generated in summer, when the vegetation is in an advanced
period of growth.
The use of prior knowledge plays an important role, which is
represented in the semantic model with an additional GIS-layer.
Vector data of the ATKIS DLMBasis (German Authoritative
Topographic-Cartographic Information System) is used, which
is an object based digital landscape model of the whole country:
(1) Field boundaries and wind erosion obstacles are exclusively
located in the open landscape, thus, further investigations are
focused to this area. The open landscape is not directly
modelled in the ATKIS DLMBasis, this is why this information
has to be derived by selecting all areas, which are not
settlements, forests or water bodies. (2) The road network,
rivers and railways can be used within the open landscape as
prior knowledge: The geometries of the GIS-objects are
introduced in the semantic model with a direct relation from the
GIS-layer to the real world (cf. Figure 1) For example, the
ATKIS objects 3101 (road) and 3102 (path) are linked to the
road segment of the real world and, thus, are usable as field
boundaries (e.g. a road is a field boundary). The modelling of
the GIS-objects in the geometry and material layer together
with the image layer is not of interest, because they do not have
to be extracted from the imagery (depicted with dashed lines in
Figure 1). (3) Tree rows and hedges are only captured in the
ATKIS-data, if they are longer than 200 m and lie along roads
or are formative for the landscape. Therefore, the relation from
the GIS-layer to the real world is limited and a modelling in the
image layer is also required.
The first object to be extracted, the field, is divided in the
semantic model in field area and field boundary in order to
allow for different modelling in the layers: The field area is a
2D vegetation region, which is a homogeneous region with a
high NDVI (Normalized Difference Vegetation Index) value in
the CIR-image. The field boundary is a 2D elongated vegetation
boundary, which is formed as a straight line or edge in the
image. Both descriptions lead to the desired result from
ditferent sides.
The second object to be extracted, the wind erosion obstacle, is
divided in hedge and tree row due to different available
information from the GIS-layer. Afterwards, both objects are
merged, because modelling in the subsequent layers is identical.
In addition, the wind erosion obstacles are not only described
by their direct appearance in geometry and material, but also
with the fact, that due to their height (3D object) there is a 2D
elongated shadow region next to and in a known direction (e.g.
northern direction at noon). Therefore, the "concrete of"
relation not only consists of an elongated, coloured region with
a high NDVI value, but additionally of an elongated dark region
with a low NDVI value alongside the real object of interest in a
known direction.
The integrated modelling leads in particular to relationships
between the field boundaries and wind erosion obstacles, thus,
inside the layers of the semantic model (cf. Figure 1): One
object can be part of another one or be parallel and nearby, and
together they form a network in the real world. For example,
wind erosion obstacles are not located in the middle of a field,
because of disadvantageous cultivation conditions, but solely on
| Landscape |
Open
Landscape
| | |
river ROO T ^ road - [ sedi | wind erosion
| network | obstacle j
, 1
J]i ania M Rs
/ : junction railway- | road | field boundary | field area | | hedge [ treerow |
| |
Real / : segment J| segment av a es Ÿ Pt rare ee —
World / LL em me a uL mm erm em um eo TA Ren ss :
i isa/is paralig with a short distance in bgween — ,
Amm mm nf m m] sn mm | i Re ec et te Ten si mm ra cme fn Fm re ce ss me] em: a N re ne am some J’ tn om
, : :
GIS- ATKIS-object ATKIS-object ||| ATKIS-object
layer 5101, 5102, 510 | = a0 ^ 4202
Geometry CoD ejongaled: | 2D compact |: 2D elongaled.
d Material : water Sent bal
and Materia P aedem. ek een DI eden ford, dede! |. Te,
i" elongated "compad ^: elongated | ’compad {elongated |
Image (CIR) dark region, : dark region, | derkregion, i ! light/dark :; light/dark
ow NDVI : low NDVI i: low NDVI | 1 region low :! region, low '
ee nes voue 31 vale |} : NOM vae ii NDVI value;
legend ;
“ ; ^ , te à” 3 special
part of concrete of is a ns GIS-database M 3
relation relation ? relation relation > ee
Figurel. Semantic Model
1066
In
th
di
di
n
pr
