FEATURE BASED IMAGE ORIENTATION AND OBJECT RECONSTRUCTION
Günter Strunz
Chair for Photogrammetry and Remote Sensing
Technical University Munich
Arcisstraße 21, D-8000 Munich 2
Germany
Commission III
ABSTRACT:
Image orientation and object reconstruction tasks in photogrammetry are generally based on distinct points.
However, in 3D object space besides points there are also lines (straight or curved) and surfaces (planar or curved),
which are increasingly being available in digital data bases, and in 2D image space lines can be extracted, especially
when automatic procedures of image analysis are used. These geometric features in object space and their
projections into image space can be used for image orientation and object reconstruction.
In this paper a systematic overview concerning the use of points, lines and surfaces for photogrammetric orientation
and object reconstruction tasks is given. It is shown, which tasks can be performed, and the respective minimum
number of features needed for a solution is derived. Some remarks on degenerate configurations are given and the
28
evaluation of precision using general geometric features is shown by means of simulated examples.
KEY WORDS: Image orientation, general geometric features, feature based photogrammetry.
1. INTRODUCTION
The transition from analytical to digital photo-
grammetry, which is one of the major topics in
photogrammetric research, requires the modification
and extension of the existing models and the develop-
ment of completely new concepts. For the geometric
tasks of photogrammetry this basically means that the
photogrammetric observations in image space do not
result from human measurements, but from automatic
techniques of image analysis. In the traditional
approach image orientation and object reconstruction
tasks are more or less completely based on distinct
points. However, geometric features in object space
can also be straight or curved lines and planar or
curved surfaces. (Note: the term lines refers to one-
dimensional curves including both straight and curved
continuous features, i.e. it is not restricted to straight
lines). These features and their projection into image
space can be used for image orientation and object
reconstruction. Especially for digital techniques of
image analysis this aspect is essential, because lines
are easier to extract than point features by automatic
procedures. Therefore a sound theory is necessary for
dealing with general geometric features.
Research and practical work based on the correspon-
dence of general geometric features, besides points, is
being performed in computer vision and photo-
grammetry. Some references are given in the follow-
ing, which represent important work done in this field.
113
Horizontal and vertical lines were used as control
features for close range applications in Dóhler (1975).
Kager, Kraus (1976) described the mathematical
model for plumb lines, straight lines, parallel and
orthogonal lines and planes in a simultaneous
adjustment of photogrammetric, geodetic and so-
called fictitious observations. The concept of line to
line correspondence in aerial photogrammetry was
shown by Masry (1981) and Lugnani (1982). The
rectification of SPOT imagery using points and edges
as control was presented by Paderes et al. (1984).
Space resection based on straight line correspon-
dences was shown in Dhome et al. (1989), Liu et al.
(1990) and Chen (1991). The formulation of the
mathematical model for the solution of space resec-
tion and intersection using line features was given in
Mulawa, Mikhail (1988); further work was published
in Sayed, Mikhail (1990). Relative orientation of a
sequence of images using correspondences of straight
line features was treated e.g. in Yen, Huang (1983),
Liu, Huang (1988a, 1988b), Spetsakis, Aloimonos
(1990) and Weng et al. (1992). In these approaches
the computation of the rotation and position para-
meters is separated and non-linear as well as linear
algorithms are derived. In Kubik (1991) relative and
absolute orientation using linear features was dis-
cussed. Finsterwalder (1991) presented algorithms and
geometric explanations for the use of lines in space
resection, rectification and absolute orientation. 3D to
3D matching of corresponding lines and surfaces was
published by Faugeras, Hebert (1986), Kim, Aggarwal