Peer Reviewed Articles
769 Mapping by Dragging and Fitting of Wire-Frame Models
George Vosselman and Henri Veldhuis
Abstract
Download
Full Article
Semi-automatic measurement of objects with regular shapes
can be performed efficiently in three steps: (1) selection of an
object model and approximate alignment of its wire frame by
an image analyst, (2) precise alignment to the image by a
fitting algorithm, and (3) correction of fitting errors, again by
the image analyst. This paper presents a new approach to
perform these three steps using the same principle in all three
steps. The developed approach allows the image analyst to
drag both points and lines of the projected wire frame,
including curved edges and contour edges, in order to align
these features with the image contents. Using the described
algorithm, there is no need for the image analyst to specify
which parameters of the object models are to be adapted in
order to improve the alignment. The performance of the fitting
step is analyzed and compared with an alternative approach.
777 Automatic Road Extraction Based on Multi-Scale, Grouping,
and Context
Albert Baumgartner, Carsten Steger, Helmut Mayer, Wolfgang Eckstein,
and Heinrich Ebner
Abstract
Download
Full Article
An approach for the automatic extraction of roads from digital
aerial imagery is proposed. It makes use of several versions
of the same aerial image with different resolutions. Roads are
modeled as a network of intersections and links between these
intersections, and are found by a grouping process. The context
of roads is hierarchically structured into a global and a local
level, The automatic segmentation of the aerial image into
different global contexts, i.e., rural, forest. and urban area, is
used to focus the extraction to the most promising regions. For
the actual extraction of the roads, edges are extracted in the
original high resolution image (0.2 to 0.5 m) and lines are
extracted in an image of reduced resolution. Using both resolution
levels and explicit knowledge about roads, hypotheses
for road segments are generated. They are grouped iteratively
into larger segments. In addition to the grouping algorithms,
knowledge about the local context, e.g., shadows cast by a
tree onto a road segment, is used to bridge gaps. To construct
the road network, finally intersections are extracted. Examples
and results of an evaluation based on manually plotted
reference data are given, indicating the potential of the
approach.
787 Virtual City Models from Laser Altimeter and 2D Map Data
Norbert Haala and Claus Brenner
Abstract
Download
Full Article
Virtual reality applications in the context of urban planning
presume the acquisition of three-dimensional (3D) urban
models. Photo realism can be achieved only if the geometry of
buildings is represented by a detailed and accurate CAD model
and if artificial texture or real world imagery is mapped to the
faces and roofs of the buildings. In the approach presented
in this paper, height data provided by airborne loser scanning
and existing ground plans of buildings are combined in order
to enable on automatic 3D data capture. On demand, automatic
building reconstruction can be visually controlled and refined
by an interactive tool. Virtual reality city models are generated
in the find step by mapping terrestrial images to the facades
of the reconstructed buildings. Thus, the rapid acquisition of
a 3D urban GIS is feasible.
797 Objects with Fuzzy Spatial Extent
Tao Cheng and Martien Molenaar
Abstract
Download
Full Article
The determination of the spatial extent of geo-objects is
generally approached through their boundaries or, more
precisely, through the positions of their boundary points. The
analysis of the geometric uncertainty of the objects is therefore
often based on accuracy models for the coordinates of these
points. In many survey disciplines objects are mapped,
however, that are not crisp well defined. In that case, the
geometric uncertainty is not only a matter of coordinate
accuracy, but also a problem of object definition and thematic
vagueness. The spatial uncertainty of such objects cannot be
handled by a geometric approach alone, such as the epsilon
band method. This paper investigates the reasons for the fuzzy
spatial extent of objects and proposes an approach to map the
spatial extent of objects and their uncertainties when objects
are extracted from field observation data. The relationship of
uncertainties between thematic aspects and geometric aspects
is investigated. A practical example of a coastal geomorphology
study is discussed to illustrate the approach.
803 Automatic Matching of Buildings and Corners
Han-Wen Hsiao and Kam W. Wong
Abstract
Download
Full Article
To update a portion of an existing cartographic database,
the common practice is to relate a new data file to an existing
file by means of survey control points that are included
in both files. In the absence of such survey control points,
well-defined points such as building corners can be used.
This paper presents an algorithm to perform matching of
common buildings and building corners in vector data files.
The algorithm starts with a Fourier-based initial matching. A
sequence of validity checks combined with robust estimation
provides a complete recognition of common buildings. Matching
of individual corner points is performed by using a similarity
parameter, followed by a series of checks and validations.
The two maps may have different scales. different
coordinate system, and-no identifying cartographic labels.
Experimental results have demonstrated the robustness of
the algorithm.
811 Knowledge-Based Interpretation of Remote Sensing Images Using Semantic
Nets
R. Tönjes, S. Growe, J. Bückner, and C.-E. Liedtke
Abstract
Download
Full Article
The increasing amount of remotely sensed imagery requires
efficient analysis techniques. The leading idea of the presented
work is to automate the interpretation of aerial images
by the use of common a priori knowledge about landscape
scenes. In addition, the system uses specific map knowledge
of a GIS. The a priori knowledge about landscape scenes, the
aerial images, and the image forming sensors is represented
explicitly by a semantic net. The definition of a network
language allows the exploition of the knowledge base by a set
of application-independent rules which provide data and
model-driven control strategies. Competing interpretations are
stored in a search tree and judged considering their uncertainty
and imprecision. An A*-algorithm selects the most
promising interpretation for further analysis. Results are
shown for the extraction of roads and complex objects, such
as purification plants, from multisensor imagery.
823 Extracting 3D Information Using Spatio-Temporal Analysis
of Aerial Image Sequences
Guoqing Zhou
Abstract
Download
Full Article
To overcome depth discontinuities end occlusion problems in
three-dimensional (3D) surface information extraction using
traditional stereophotogrammetric matching, a new approach
called spatio-temporal analysis of aerial image sequences is
proposed. In the proposed methad, a set of spatio-temporal
solid data is first formed from a sufficiently dense sequence of
images taken by a camera moving along a straight-line path.
Second, the set of spatio-temporal solid of data is sliced
along a temporal dimension in to epipolar-plane images
(EPIs), and features in these slices are extracted and described.
Finally, three-dimensional coordinates in a ground
coordinate system are computed for the features in the EPIs.
This method is fairly radically different from traditional two view
stereophotogrammetric matching; therefore, we discuss
in detail the estimation accuracy, error resources, and sensitivities
to occlusion and depth discontinuities. The experimental
results from three test fields in Berlin, Germany show
that the method is a useful tool for solving the problems of
depth discontinuities and occlusion with which photogrammetrists
have been wrestling for a decade.
| Top | Home |