Geography Reference
In-Depth Information
provide tools for graphic-oriented tasks but rather tools for data model
(non-graphic) generalization. Model-oriented generalization takes place within
the scope of an internal representation of a map and pursues reduction of the
information density in a database (Filippovska et al.
2008
). Contemporary trend
of research is focused on automated methods that enable the creation and display of
geographic information at multiple levels of detail (Cheng and Li
2013
).
The greatest need of the generalization feel organizations that operate large
volumes of spatial data (such as National Mapping Agencies). Their requirements
for automated generalization are still not directly available. The reason for these
missing requirements are that: (a) currently no formalism has proven to be adequate
for fully capturing the specifications of a map, (b) not all requirements are easily to
be formalized and (c) much knowledge on generalization requirements and pro-
cesses still needs to be revealed (Stoter et al.
2009
).
Although principles and guidelines of the generalization can be found in carto-
graphic literature and among mapping organizations, there has not existed a set of
universal rules that explicitly and completely defines how the generalization should
be performed (ESRI
1996
).
The challenge of the generalization is to preserve the geographical meaning and
relations; it is necessary to recognize between essential and unimportant informa-
tion (Bard and Ruas
2005
). The generalization is a subjective process with an accent
to knowledge and experience of the cartographer. The computational geometry
makes a process of the simplification less dependent on a subjective view of the
cartographer, however it is not an easy task to find and set a geometric criterion that
should be satisfied by a simplified element (Bayer
2009
).
Due to this fact few authors were searching for approaches or tools how to
evaluate results of generalization using various methods (e. g. Girres
2011
;
Filippovska et al.
2008
). Shape metrics may serve as such a tool or method for
quantitative evaluation of generalization methods and their outputs.
Shape metrics were originally applied in the landscape ecology in order to
quantify characteristics of landscape patches. Landscape patches represent homog-
enous, further indivisible units of a landscape (according to the scale—forests,
humid areas, specific habitats, urban fabric, etc.). These patches are represented as
polygons in GIS and therefore are objects of cartographic generalization. Applica-
tions of shape metrics in cartography with the focus on polygons can be found in
Peter (
2001
), Schmid (
2008
), Agent (
1999
), Burghardt and Steiniger (
2005
) or Gao
et al. (
2012
). Complexity measures of generalized line features are thoroughly
described in Jasinski (
1990
), Bernhardt (
1992
), or Skopeliti and Lysandros (
2001
).
Generally, shape metrics serve as a quantitative description of any planar object
(e.g. ground projection of a building) in order to measure its shape complexity,
roughness or irregularity. Shape metrics are fundamentally based on an area of a
shape and its perimeter (these two characteristics are themselves considered as
shape metrics and are very easy to obtain), but most of the metrics are more
complicated to calculate and are treated as shape indexes. Nevertheless, it is
worth to mention, why it is useful to calculate shape metrics. Since shape metrics
take into account only geometric properties of the patch (polygon), it is possible to
