Full text: Fusion of sensor data, knowledge sources and algorithms for extraction and classification of topographic objects

International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 7-4-3 W6, Valladolid, Spain, 3-4 June, 1999 
4. Investigation and development of techniques for geocoded 
multisensor data fusion; 
5. Use of image analysis techniques to extract height 
information for 2D databases; 
6. Integration of image analysis techniques with GIS for 
querying, analysis and representation of spatial data; 
7. Treatment of uncertainties, generalisation, and scale and 
temporal differences of GIS and image derived data. 
A topic, which is not included in the above ToR but 
increasingly attracts the attention of scientists, is the integration 
of various cues, and of different algorithms and their partial 
results for object recognition and reconstruction. Among the 
ToR, the major activities of the WG and its 90 members since 
1996 have been primarily on topics 1, 2, 4 and less 5. Topic 4 
has been treated extensively in many papers of these 
proceedings (for an overview and comparison of various 
methods see Pohl, 1999; Hill et al., 1999), so our focus in this 
paper will be on the remaining 3 topics. We will concentrate on 
aerial and spacebome sensors and as tasks, classification, 
identification and reconstruction of topographic objects both in 
2-D and 3-D, using semi-automated and automated methods. 
A questionnaire was sent to our WG members in 1998 to collect 
information on their research activities, publications, 
applications, interests etc. More than 100 references were 
received. The above information was appended by an 
additional, not complete, search of the authors and will be 
presented here. Other work on similar and related topics is 
performed by additional ISPRS WGs (e.g. H/2, II/6, III/3, III/4, 
III/5, IV/2, IV/3, VII/4; information on these WGs can be found 
at www.geod.ethz.ch/isprs), the Special Interest Group Data 
Fusion’ (www-datafusion.cma.fr/sig/) of EARSeL and the 
proceedings of the associated "Data Fusion" conferences, and 
some OEEPE WGs (www.itc.nl/~oeepe), e.g. the WG on 
Automatic Absolute Orientation on Database Information 
(www.i4.auc.dk/jh/OEEPEgroup.htm). The responses showed 
that the developments in this field, although continuously 
increasing, are quite fragmented and in various heterogeneous 
fields and applications. A clear overview of these developments 
and underlying unifying theories is missing. The expectations 
and the interest from people involved in production were high. 
A representative answer was "As a National Mapping Agency 
...very interested..., especially if it led to systems which we 
could use in our day-to-day activities of maintaining geospatial 
The major applications addressed were the following: 
• Fusion of panchromatic and spectral data (SPOT, IRS, 
Landsat, MOMS, DP A); often used for improvement of visual 
interpretation in application-specific environments, e.g. in 
forestry or military reconnaissance (Zhukov et al., 1995; 
Garguet-Duport et al., 1996; Wald et al., 1997; Steinnocher, 
1999; Pohl and Touron, 1999). 
• Fusion of optical images and SAR (e.g. hybrid orthoimages) 
(Pohl, 1996) or of products derived from them, like DTMs 
(Honikel, 1999). 
• Use of GIS/maps for automatic DTM generation, image 
segmentation and object extraction (especially buildings, 
roads, landcover classes), and generation of 3-D city models 
(van Cleynenbreugel et al., 1990; Janssen et al., 1990; Solberg 
et al., 1993; Maitre et al., 1995; Quint and Sties, 1995; de 
Gunst, 1996; Quint and Landes, 1996; Roux et al., 1996; 
Roux and Maitre, 1997; Bordes et al. 1997; Haala et al., 1997; 
Huang and Jensen, 1997; Koch et al., 1997; Stilla et al., 1997; 
Schilling and Vogtle, 1997; Tonjes, 1997; Quint, 1997a, 
1997b; Prechtel and Bringman, 1998; Zhang, 1998; Stilla and 
Jurkiewicz, 1999). 
• Integration of image and map data in GIS (e.g. for a forest 
information system) (Dees and Koch, 1997). 
• Use of GIS for automatic training area selection and 
verification of the results in landcover and landuse 
classification, especially in agriculture and forestry (Walter 
and Fritsch, 1998). 
• Use of ortho- or normal images in GIS for updating of 
topographic or thematic maps (Duplaquet and Cubero-Castan, 
1994; Newton et al., 1994; Plietker, 1994; Aas et al., 1997; 
Vosselman and de Gunst, 1997; Duhaime et al., 1997; Peled 
and Haj-Yehia, 1998; Walter and Fritsch, 1998). Usually the 
updating is done manually using orthoimages as a backdrop, 
but steps towards automation in national mapping 
organisations (Israel, IGN, Canada planned) have been 
• Use of GCPs from maps/digital databases and their detection 
in images, or use of orthoimages and DTMs for automatic 
image orientation and geocoding (Pedersen, 1997; Drewniok 
and Rohr, 1997; Hoehle, 1998, Sester et al., 1998); related to 
the topic below. 
• Automatic registration of images to images, (vector) maps and 
models (Roux, 1996; Growe and Tonjes, 1997; Ely and Di 
Girolamo, 1997; Dowman and Ruskoné, 1997; Vasileisky and 
Berger, 1998; Dowman and Dare, 1999). 
• Registration of images to site-models (similar to previous 
topic, but related more to change detection, often in the 
context of military applications) (Chellapa et al., 1994; 
Mueller and Olson, 1995; Huertas et al., 1995). 
• Use of image analysis for automatic interpretation and 
vectorisation of maps (Frischknecht et al., 1998). 
• Use of image analysis in data mining, image retrieval and 
queries (Agouris et al., 1998). 
• Matching of maps and vector datasets (often termed 
"conflation"), e.g. for combination of one road vector dataset 
with good geometry with another one having poor geometry 
but rich and up-to-date attributes. 
• Use of image analysis to extend existing spatial databases 
from 2D to 3D (Axelsson, 1997; Lammi, 1997). 
• Combination of different cues (indicators) for classification, 
object recognition and reconstruction (multi- and hyper- 
spectral properties, texture, 3D form from DSMs and DTMs 
derived from imagery or laser scanners, morphology and 
shape, shadows etc.) (Solberg et al., 1994, 1996; Moissinac et 
al., 1995; Strat, 1995; Henricsson et al., 1996; Baltsavias and 
Mason, 1997; Baumgartner et al., 1997; Bruzzone et al., 
1997; Stolle et al., 1997; Lemmens et al., 1997; Piesbergen 
and Haefner, 1997; Hahn and Statter, 1998; Csathó et al., 
1999; Haala and Walter, 1999).

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.