developing automated methods to partition remote sensing imagery into meaningful
image-objects (i.e., patches), and assessing their characteristics through scale.” (Hay
and Castilla, 2008, p. 79).
8.1.2 Geographic Object-Based Hierarchies
Hierarchies are granular partitions (i.e., parts composed of parts), and thus from a
geographical or landscape perspective, they can be divided into either taxonomies
(representing partitions of the conceptual domain) or zonations (partitions resulting
from projecting the taxonomy onto the territory) (Castilla, 2003). This important
distinction is compatible with the object-oriented (OO) paradigm as defined in
computer science. In this paradigm, two kinds of hierarchical structures are recognized, each of which describes different types of relationships between classes
(Graham, 2001). A class refers to a group of objects with similar attributes and
behaviors, and an object is any physical or conceptual entity. The first type of OO
hierarchy describes the aggregation relationship by asking if a focal object (e.g., tree)
is a part of a higher level object (e.g., forest stand), or reciprocally if it is composed of
a lower level object (e.g., bole, branches, and leaves). True responses to both
questions typically result in a nested hierarchy (Figure 8.1).
The second type of OO hierarchy describes the generalization–specialization
relationship and asks whether an object at the lower levels is a kind of focal object.
This results in generalization: For example, a pine is a kind of tree. Conversely, the
upper level class can be the object at the focal level. For example, vegetation can be a
tree, grass, or flowers, which represents specialization. True responses typically result
in both nested and unseated hierarchies. For example, in Figure 8.2, pine and tree are
each a kind of class above, while vegetation and tree can be the class beneath, and thus
they are hierarchically nested. Similarly, maple and flowers are a kind of vegetation;
however, both cannot be a kind of tree. Thus, flowers are unseated with regards to tree
but nested with regards to vegetation. The important point to appreciate is that while
both types of hierarchies may contain the same focal object (i.e., tree), their
FIGURE 8.1 Hierarchical structures showing object-oriented aggregation relationship.
INTRODUCTION
145
image-objects (i.e., patches), and assessing their characteristics through scale.” (Hay
and Castilla, 2008, p. 79).
8.1.2 Geographic Object-Based Hierarchies
Hierarchies are granular partitions (i.e., parts composed of parts), and thus from a
geographical or landscape perspective, they can be divided into either taxonomies
(representing partitions of the conceptual domain) or zonations (partitions resulting
from projecting the taxonomy onto the territory) (Castilla, 2003). This important
distinction is compatible with the object-oriented (OO) paradigm as defined in
computer science. In this paradigm, two kinds of hierarchical structures are recognized, each of which describes different types of relationships between classes
(Graham, 2001). A class refers to a group of objects with similar attributes and
behaviors, and an object is any physical or conceptual entity. The first type of OO
hierarchy describes the aggregation relationship by asking if a focal object (e.g., tree)
is a part of a higher level object (e.g., forest stand), or reciprocally if it is composed of
a lower level object (e.g., bole, branches, and leaves). True responses to both
questions typically result in a nested hierarchy (Figure 8.1).
The second type of OO hierarchy describes the generalization–specialization
relationship and asks whether an object at the lower levels is a kind of focal object.
This results in generalization: For example, a pine is a kind of tree. Conversely, the
upper level class can be the object at the focal level. For example, vegetation can be a
tree, grass, or flowers, which represents specialization. True responses typically result
in both nested and unseated hierarchies. For example, in Figure 8.2, pine and tree are
each a kind of class above, while vegetation and tree can be the class beneath, and thus
they are hierarchically nested. Similarly, maple and flowers are a kind of vegetation;
however, both cannot be a kind of tree. Thus, flowers are unseated with regards to tree
but nested with regards to vegetation. The important point to appreciate is that while
both types of hierarchies may contain the same focal object (i.e., tree), their
FIGURE 8.1 Hierarchical structures showing object-oriented aggregation relationship.
INTRODUCTION
145