Je B € a) Ad
UE TS m
no ——
interests of this Working Group is rather broad. But one
could broaden it even further if one were to consider sur-
faces from point clouds irrespective of their origin. This
would include point clouds obtained from images, from
underwater SONAR, from profiling techniques.
The Working Group on automated object modelling has
three predecessors in the 1996-2000 period At the time it
was argued that the subject matter should be separated
into some form of low-level, mid-level and high-level
vision. We are abandoning this separation at this time in
the WG-structure. Obviously, this represents the topic of
'automated image interpretation' . Considerable interest
exists in photogrammetry since this has applications in the
automated population of geographic data bases. Topo-
graphic objects such as roads, buildings, fences, bridges
etc. need to be mapped. The question immediately comes
to mind: ‘Is there a photogrammetric automated object
recognition?’ Is it ‘photogrammetry’ when the objects are
topographical? Probably not. But the question illustrates
that we need to create an understanding where pho-
togrammetry stands, and this Working Group can and
should help in achieving this clarity.
A new Working Group on the Theory and Algorithms for
industrial vision is being introduced, consistent with the
Commission’s charter to address and focus on theory and
algorithms. A careful co-operation with Commission V is
needed in this area. The new Working Group is interested in
3-D vision in industrial settings, in reconfigurable calibration
and, most importantly, in the hot topic on ‘uncalibrated
vision’, a concept that is counterintuitive to photogram-
metrists, but makes a lot of sense in the proper context.
The Working Group on Fusion has a legacy in the preced-
ing period. One might argue that ‘fusion’ of data, data
structures and methods is everywhere, and therefore
should not be a separate focus. But by having a separate
working group on conceptual aspects of information fusion,
or multi source vision, one demonstrates that the basic
necessity of using multiplicity, where available and reason-
able, needs more attention than it currently receives. InSAR
is part of an integrated bundle of data including SAR-
coherence, magnitude, polarisation and shape, and there-
fore aspects of InSAR will find a home in this WG.
Another new Working Group is to address Virtual Environ-
ments. Both Virtual Reality and Augmented Reality create
a need for rapid modelling of the human habitat and envi-
ronment. This issue is of course present in many Working
Groups, even Commissions. But as far as ‘Automation’ is
concerned, Commission Ill needs to become active, not
only in the context WG on automated object modelling.
Issues are the integrated analysis of both terrestrial and
aerial imagery, the extraction of texture, automated gener-
ation of models with level-of-detail, the inferred attribution
Rendering in real time and following a moving user pres-
ents its own complexities. This implies that tracking be
very accurate and in real time. City modelling is the most
often discussed application for photogrammetric Virtual
Environments.
Very important is the concern for reliability and perform-
ance of algorithms, as reflected in a Working Group for this
topic. The subject is or should be ubiquitous. But by cre-
International Archives of Photogrammerty and Remote Sensing. Vol. XXXIII, Part A. Amsterdam 2000.
ISPRS
— 20
ating a separate working group, we 'flag' the topic and
expect it to create guidelines, test data, ideas, references
and algorithms for the use by others in assessing the value
of their creative algorithm work.
In order to better consider the world of computer vision, it is
recommended that we seek to ‘populate’ computer vision
conferences. Working Groups will organise workshops
before, after or as part of those conferences. In this manner,
there is reinforced attention being paid to the events and
innovations in those vision conferences, and people attend-
ing those conferences learn better to appreciate what pho-
togrammetry is. Again, the general motto could be ‘Pho-
togrammetric Computer Vision’, and present session under
this topic at EVVC, ICCV, CVPR, ICOPR, CAIP and the likes.
This can be as illuminating for photogrammetrists as it can
be good marketing in the vision communities.
Working Groups of Technical Commission lll for 2000-
2004
WG IIl/1 Sensor Pose Estimation
Chair: Henrik Haggrén (Finland)
Co-Chair: Ayman Habib (USA)
WG IIl/1 Terms of Reference
- Block adjustment: projective vs. perspective transfor-
mation
- Registration algorithms
- Orientation procedures for 3-D scene reconstruction
- Block triangulation for airborne digital sensors and
cameras
- Use of features as entities in image orientation
processes
WG IIl/2 Surface Reconstruction from Images As
Information Source
Chair: Michel Roux (France)
Co-Chair: Amnon Krupnik (Israel)
WG IIl/2 Terms of Reference
- Stereo matching
- Shape from X
- 3-D versus 2.5D (in collaboration with WG III/3);
- Improvement provided by automated object identifica-
tion and by image/scene understanding (in collabora-
tion with WG IlI/4)
WG IIl/3 3-D Reconstruction from Airborne Laser
Scanner and InSAR Data
Chair: George Vosselman (The Netherlands)
Co-Chair: Hans-Gerd Maas (Germany)
WG IIl/3 Terms of Reference
- Algorithms for point cloud processing (in collaboration
with WG IIl/2 on surface reconstruction)
- Data fusion (in collaboration with WG 11/6)
- Products: Digital surface models, digital elevation
models, 3-D city and landscape models
- Applications: (in collaboration with WGs VII/3, VII/4,
VII/5 on coastal mapping, flood prediction, urban plan-
ning, telecommunications planning, monitoring of
power lines, noise and gas propagation, tax verifica-
tion, real estate sales etc.)
