Pakzad, Kian
2 KNOWLEDGE BASED INTERPRETATION SYSTEM
The system used for the presented approach is the knowledge based system AIDA (Liedtke et. al. 1997, Toenjes 1998)
which was developed in order to automatically interpret remote sensing images. The system strictly separates the
control of the image analysis process from the semantics of the scene. The knowledge representation is based on
semantic nets (Niemann et. al. (1990)). Semantic nets are directed acyclic graphs. They consist of nodes and edges in
between the nodes. The nodes represent the objects expected in the scene while the edges or links of the semantic net
model the relations between these objects. Attributes define the properties and methods of nodes and edges.
The nodes of the semantic net model the objects of the scene and their representation in the image. Two classes of
nodes are distinguished: the concepts are generic models of the object and the instances are realizations of their
corresponding concepts in the observed scene. Thus, the knowledge base which is defined prior to the image analysis is
composed of concepts. During the interpretation a symbolic scene description is generated consisting of instances.
The relations between the objects are described by edges or links of the semantic net. The specialization of objects is
described by the is-a relation introducing the property of inheritance. Along the is-a link the description of the parent
concept is inherited to the more special node which can be overwritten locally. Objects are composed of parts
represented by the part-of link. Thus, the detection of an object can be simplified to the detection of its parts. The
transformation of an abstract description into its more concrete representation in the data is modelled by the concrete-of
relation, abbreviated con-of. This relation allows for structuring the knowledge in different conceptual layers, for
example a scene layer and an image layer.
To make use of the knowledge represented in the semantic net control knowledge is required which states how and in
which order the image interpretation has to proceed. The control knowledge is represented explicitly by a set of rules.
The rule for instantiation for example changes the state of an instance from hypothesis to complete instance, if all
subnodes which are defined as obligatory in the concept net have been completely instantiated. If an obligatory subnode
could not be detected, the parent node becomes a missing instance. The control of interpretation is also performed by an
A*-Algorithm. For further details see Toenjes (1998).
For the interpretation of multitemporal images the system was extended by temporal relations. They realize the use of
temporal knowledge which is described in state transition diagrams (see section 5.4). For each temporal relation a
priority can be defined in order to sort the possible successor states by decreasing probability. As states can either be
stable or transient, the corresponding state transitions differ in their transition time which can be also specified in the
temporal relation. During scene analysis the state transition diagram is used to generate hypotheses for the next
observation epoch. For each of these possible state transitions a hypothesis is generated. All hypotheses are treated as
competing alternatives.
3 PRIOR KNOWLEDGE
In this section an overview of the prior knowledge about moorland is given (see Redslob (1999), Eigner (1991),Gôttlich
(1990) for further details). This knowledge will be implemented in the knowledge base for the monotemporal
interpretation (see section 5.3) and the multitemporal one (see section 5.4). With the example moorland interpretation
the way from the general knowledge described below to the explicit representation form which is used in our approach
and which can be processed by the knowledge based system is demonstrated.
Originally, moors were upland moors. In Germany these have practically vanished. Today mostly agriculturally used
areas, forests and areas of regeneration or degeneration are found in the former upland moors. The most important
industrial use of moorland is the peat extraction. In order to make peat extraction possible in a moor the ground has first
to be drained. Therefore ditches need to be created. Thus, the water level goes down and the area begins to degenerate.
The vegetation changes. During the state of degeneration the vegetation is inhomogeneous and irregular. Then, peat
extraction is possible. Usually harvester machines are used. In aerial images the use of the machines can be recognized
by parallel tracks. After peat works have finished, a regeneration of the moorlands can begin. In most cases people
simply stop working the land and leave it to regenerate, which eventually results in increasing vegetation. Hence, in this
state of land use vegetation can be found again on these areas, especially birches because of the dry ground. In many
cases also remains of the tracks from the harvester machines from the state before can still be found. In order to start up
a regeneration in the direction of the original state, upland moor, sometimes supporting steps are being carried out. Such
steps can be to fill up the ditches und to remove trees in order to reach a raising water level, which is one prerequisite
for the goal. If the water level does rise trees out die and a homogeneous vegetation without trees grows.
1104 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.
FN FM IS om 7 OR