If the structure feature represent a valley (positive curvature),
the adaptive filter method lead to a less height of the DEM
posts in comparison to the global filter method. If the structure
feature correspond to a ridge (negative curvature) it is vice
versa.
ha ^9 (positive curvature)
hy <0 (negative curvature)
In flat terrain the height differences between the globally fil-
tered DEM and the adaptively filtered DEM are negligible.
A
=
ZT
SER
OO ER M e RS
CSS ERN EDS
= NS NS
NS SNS
Figure 4.2: Globally filtered DEM
SITZ
ZT ve
ve =
cs.
EN
> sn.
N NECS =
RS
NN ER
J ZN
ess
os
SSSR SS y
X \ >
ENG
: y
^
Me
AVANCE
LANE
D. x
Figure 4.4: Differential DEM
The quantitative difference is expressed by the r.m.s. height
difference and amounts to
iE
tus =, Ln — 0.64 m
d= height differences of the DEM posts
n- number of DEM posts
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
Finally we conclude that, the use of the adaptive edge preserv-
ing filter takes effect in areas of breaklines not parallel to the
X- and Y-direction of the object coordinate system. Because
of this the recognition of small surface undulations reacts more
sensible leading to a better success rate in automatic breakline
detection and grid width adaptation.
4.3 Derivation of structure information
The first step for the automatic breakline detection is the der-
ivation of structure information (approximate breakline points
and corresponding local breakline directions) in the object
space during the DEM generation process. The implemented
adaptive edge preserving filter provides this information. For
that reason we use this surface reconstruction method to deal
with the following data.
A 4 E ; dd
Figure 4.5: Open coal-mine excavation area
Xs, ox, t v c oS,
E
Figure 4.6: Mountainous terrain
ar
to
th
