flying height, flying speed and scanner frequency (Lemmens ef
al., 1997). Characteristics and performance of laser data systems
have been discussed by many researchers (Ackermann, 1999;
Axelsson, 1998; Baltsavias, 1999; Fritsch, 1999: Hug and Wehr,
1997; Kilian et al., 1996; Lohr, 1998; Vaughn er al., 1996).
Calibration methods and the errors which may occur in the data
have also been investigated (Fritsch and Kilian, 1994; Hu er al.,
1998; Huising and Gomes Pereira, 1998; Lemmens, 1997;
Lemmens et al., 1997; Kilian, 1994). Data processing and
filtering methods have been described by Axelsson (1999), Hug
and Wehr (1997), Kilian et al. (1996) and Knabenschuh and
Petzold (1999).
Laser data consists of coordinate information only, and therefore
lacks thematic information (Ackermann, 1999; Axelsson, 1999;
Haala et al., 1997; Kraus and Pfeifer, 1998; Petzold et al., 1999).
Laser data provides accurate points with high spatial frequency,
however breaklines are not present in the data (Kraus and Pfeifer,
1998), and therefore the position of surface discontinuities can
only be estimated or calculated by methods such as segmentation
of the range data (Haala et al., 1997; Vosselman, 1999). To
illustrate this point, Figure 1 presents an elevation image showing
high-rise buildings in laser data. The edges of the buildings are
not well defined, though a high spatial density of the laser data
points is indicated by the ‘ragged’ nature of the edges, as also
noted by Vosselman (1999).
Figure 1. Plan view showing laser data.
Investigations and observations comparing DSMs produced from
laser data and those derived from digital photogrammetric
methods have been made in several research studies. In areas of
the imagery lacking texture or contrast, the image matching might
not provide accurate results whereas the accuracy of the laser is
not affected (Baltsavias, 1999; Kraus and Pfeifer, 1998). Image
matching produces a smoother DSM than the laser data at surface
discontinuities (Baltsavias, 1999; Haala, 1999; Toth and Grejner-
Brzezinska, 1999), however photogrammetric data have a higher
planimetric accuracy than laser data (Baltsavias, 1999).
The complementary nature of the two data sources has been
widely recognized and the approach of combining them has been
suggested by researchers for several years (Fritsch and Kilian,
1994; Haala, 1994). This suggestion has been reiterated recently
(Ackermann, 1999; Axelsson, 1999; Baltsavias, 1999; Brenner,
1999; Csatho er al., 1999; Fritsch, 1999; Haala, 1999; Haala and
Anders, 1997; Toth and Grejner-Brzezinska, 1999; Vosselman
International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 3W14, La Jolla, CA, 9-11 Nov. 1999
1999), The approach of using imagery to provide edge
information has been highlighted by several researchers
(Ackermann, 1999: Axelsson, 1998; Csathó et al., 1999; Haala
and Anders, 1997), and is the main concept of the approach
undertaken in this research.
The approach presented in this paper utilizes the beneficial
properties of both photogrammetric data and laser data to produce
an accurate DSM. Digital stereo imagery can provide accurate
horizontal information regarding the location of surface
discontinuities. The laser data provides accurate elevations,
however does not contain accurate location information of the
surface discontinuities, as illustrated in Figure 1. Thus the two
data sources are merged to obtain the advantages of each, and
therefore facilitating the generation of an accurate surface model.
Testing of the research approach has been undertaken using an
urban site covering Ocean City, Maryland. Laser data and aerial
images, acquired on the same day by NASA and NGS
respectively, are used. Preliminary experiments have been
performed to test and refine the algorithm. This paper presents
the surface registration and the data fusion components, describes
the data set and details the results from the initial
experimentation.
2 PROPOSED DATA INTEGRATION APPROACH
The research presented in this paper utilizes information from
laser scanner data and photogrammetric data to produce an
accurate model of the visible surface. The conceptual approach
consists of two phases. In the first phase, surfaces created from
each source are accurately registered to the same coordinate
system. The second phase consists of extracting edge information
from the photogrammetric data, which is used to delineate surface
discontinuities in the urban scene. By merging the edge
information and the laser data for surface generation, a DSM that
more closely represents the actual scene can be produced.
2.1 Surface Registration
The surface registration is undertaken to determine the
transformation parameters between the laser surface and the
photogrammetrically derived surface. Theoretically, the two data
sets should be on the same coordinate system, however the
systematic errors inherent in the laser data may introduce à
misalignment between the two surfaces, which must be
eliminated before data fusion may be performed accurately
(Kraus and Pfeifer, 1998). The misalignment has been observed
in the data sets used for the initial testing of the algorithm,
validating the incorporation of the surface registration component
into the algorithm.
The registration is performed to find the transformation
parameters between the surfaces. These parameters are used to
transform the laser data to the coordinate system of the
photogrammetric data, as these data have higher planimetric
accuracy compared to the laser data (Baltsavias, 1999). The
determined transformation parameters must be as accurate as
possible to ensure no unnecessary degradation occurs in the
accuracy of the surface generated from merging the data sets.
Internatior
The algorithm was c
surface models of
acquisition methods,
laser scanner data a
photogrammetric datz
translation parameter
factor. The surface |
and there are no ident
more detailed inforn
testing, the reader is r
2.2 Data Fusion
The main emphasis
algorithm to produce
areas, utilizing both |
The data fusion cor
stereo imagery to ob
discontinuities. The
are used as breaklinc
data. A new surface
expected to have a hi
either of the separate
The conceptual appr
that the edges 7
automatically derive
imagery using edge «
Initial research has b
of 3D line segments.
The edge pixels are
digital images. At
(Sarkar and Boyer, |
pixels. The edge pix
using these lines, st
segments are used as
and can be more ac
using feature matchi
Also, each line segn
and end points of the
For each straight lir
image is searched
searching of the ima;
to fine approach is
search process, inc
(Grewe and Kak, |
orientation parameter
segments to be det
segments is complete
The automatic extrac
must be addressed.
surface discontinuiti
foreshortening and 1:
be a number of inc
breaklines may be
discontinuities may |
refer to visual edges
