Resolution Il.6 Integration of information into GIS 
The Congress 
Noting 
the increasing need for up-to-date geospatial informa- 
tion and the lack of efficient, timely revision of such 
information in many areas 
Recognising 
that multispectral and stereoscopic imagery can provide 
such information and is becoming increasingly impor- 
tant for use in geographic information systems (GIS) 
Recommends 
that the integration of photogrammetric and remote 
sensing imagery and techniques into GIS for efficient 
acquisition and revision of geospatial information be 
strengthened. 
Resolution 11.7 End-to-end systems 
The Congress 
Noting 
the much increased use of geospatial information in all 
areas of public and commercial activity 
Recognising 
the need for efficient processing and presentation of 
such data in a value-added form 
Recommends 
the development and validation of end-to-end pro- 
cessing systems for specific applications, making use 
of a range of imaging systems, a range of components 
from the spatial information sciences and paying par- 
ticular attention to techniques for the delivery and pres- 
entation of information. 
Resolutions of Technical Commission III 
Resolution IIl.1 Surface reconstruction 
The Congress 
Noting 
the extensive use of automated surface reconstruction 
for mapping, image rectification and 3D modelling 
the emergence of laser scanning technology as an 
additional information source about surfaces 
the role of surface reconstruction in the general frame- 
work of object recognition and scene analysis 
Recognising 
the need for further theoretical investigations into the 
automatic reconstruction of surfaces, including their 
segmentation, and of conducting reliability studies 
Recommends 
that research be continued on Earth surface reconstruc- 
tion techniques with emphasis on multiple sensor input. 
Resolution III.2 Fusion 
The Congress 
Noting 
the increasing availability of new sensors and the use 
of multi-sensor, multi-resolution systems 
Recognising 
the need for extending theories and developing algo- 
rithms for merging multi-sensor data 
modelling of uncertainty in multi-sensor fusion 
incorporating GIS information to support object recog- 
nition 
evaluating the efficiency and performance of multi-sen- 
sor fusion 
International Archives of Photogrammerty and Remote Sensing. Vol. XXXIII, Part A. Amsterdam 2000. — — 
EE 2 eee Me I UT 
  
Recommends 
- that fusion, at the data, feature and information levels, 
be promoted. 
Resolution Ill.3 Object modelling 
The Congress 
Noting 
- the importance of modelling 3D objects related to 
object recognition and image understanding 
Recognising 
- that further progress in the automatic recognition of 
objects relies on improved models 
Recommends 
- that efforts be strengthened in developing generic 
models of objects, including their geometric, semantic 
and temporal properties, and interrelationships. 
Resolution IIl.4 Combining classification methods and 
computer vision 
The Congress 
Noting 
- the increased availability of multi-sensor, multi-spectral 
and hyper-spectral data 
Recognising 
- the need for combining traditional classification methods 
of remote sensing with computer vision approaches for the 
automatic recognition of objects 
Recommends 
- that efforts be strengthened in combining classification 
methodologies and computer vision approaches into a 
common object recognition framework. 
Resolution 111.5 Performance and reliability of 
algorithms 
The Congress 
Noting 
- the diversity of algorithms in photogrammetry, remote 
sensing and computer vision developed for the pur- 
pose of feature extraction and object recognition 
Recognising 
- the need for assessing the performance, reliability and 
availability of algorithms 
Recommends 
- that procedures for evaluating algorithms and for 
developing suitable test datasets be intensified and 
formulated. 
Resolution 111.6 Image understanding / object 
recognition 
The Congress 
Noting 
- the importance of theoretical and conceptual investi- 
gations in object recognition and image understanding 
Recognising 
- that despite major efforts and good progress achieved 
from 1996 to 2000 there remain considerable gaps in 
the theory for automation of feature extraction and 
recognition 
Recommends 
- that investigations in object recognition and image 
understanding be intensified, particularly in the areas 
of modelling and knowledge engineering 
- that co-operation with researchers in computer vision 
and cognitive science also be intensified. 
— 285