Geoscience Reference
In-Depth Information
Figure 2
shows an example of the segmented spectrum-line.
Then the spectrum line in the polarization space is simplified by preserving
characteristic nodes and eliminating unimportant nodes. The polarization
transformation approach is described in more detail in Qian (2005), in
Qian (2006), in Qian et al. (2006) as well as in Qian et al. (2007).
The PT-approach, however, does not predefine the number of the points
which have to be kept/eliminated for a certain map scale range. We tested
the existing PT-approach of Qian (2005) with different randomly distrib-
uted point data sets, using different amounts of point numbers. The result-
ing percentages of the eliminated points are shown in
Table 3
.
1uPber Rf pRiQts 1uPber Rf eOiPiQated pRiQts
EOiPiQated pRiQts iQ %
200
1
2
2000
18
09
5000
23
046
Table 3:
Examples of test data
What is the use for generalizing point data sets by a selection of never less
than 98 - 99,5 percent? The reason for these results lies within the empiri-
cally set angle threshold Δα in Qians PT-approach (Qian 2006).
Our enhancement aims to adapt the PT to enable point-generalizations for
user-defined scales or user-defined number of kept/selected points. Also
we enhanced the PT-approach to consider multi-dimensional point data
sets. Thereby we keep the advantage of PT, to preserve global as well as
local characteristics of spatial distributions. We improved the determina-
tion of the angle threshold Δα by defining Δα as an element of all differ-
ences of the neighbor azimuth direction angles of the spectrum line
depending on the achieved number of kept/eliminated points.
Our approach is an interactive (Web-) tool which allows to input any static
or dynamic 2D- or 3D-point data set (x,y,z,t) and generalize the points in
real time either by defining the number of points to be kept/eliminated or
by a given achieved output scale. In this paper we also provide an evalua-
tion of our point selection method.
Töpfer and Pillewizer (1966) investigated their empirical formula or “radi-
cal law” which allows determining the number of features which should
retain within a selection process. An extension of Töpfer's radical law
(principles of selection) was developed by Burghardt and Cecconi (2007)
