A. Annoni, P.M. Sargent and P.C. Smits
Space Applications Institute
Joint Research Centre (European Commission)
Ispra (VA), Italy
MOOLU-abstract2.rtf
20 April 1998 13:00
Philip Sargent
Revision History
v2 20 April 1998 13:00
v1 10 April 1998 20:00
Abstract
· Different criteria (methods) are required for interpreting images depending on whether the parcel is forest or agricultural: forest is clear-cut from year to year whereas fields may have crop rotation. To use the CORINE land-cover dataset to study both thematic and geometric change.
· Requires excellent raster/vector integration.
Introduction
Commercial object-oriented (OO) GIS software packages have achieved substantial success in cartography [Woodsford and Hardy 97] and civil utilities management [Hartnall and MacAllister 95]. This work demonstrates that OO GIS software has unique capabilities well-suited to managing common but difficult remote-sensing problems.
We show here that OO GIS enables us to solve some standard problems in land-cover interpretation [Perdigao and Annoni 97].
The key benefit that we get for image interpretation is that different image interpretation algorithms can be applied to different parts of the image, based on the classification from previous images or from externalvector information such as land parcel boundaries or other human geographic features. These algorithms are represented in the "methods" of the objects, and the objects themselves are the previous or a priori classification polygons.
Storing the algorithms as methods makes extending and changing the software much easier and less prone to error. this is because we can use the multiple-inheritance class hierarchy of land types to manage the complexity, e.g. to provide sensible default behaviour, to store any one algorithm only once, and to encode modified algorithms for special land types in a clean and controlled way.
Background
The sales department of nearly every commercial GIS has at some time claimed that their system was based on "objects" in some sense. Even though it has been five years since a reasonable scale of "object-orientedness" in GIS was published [Batty 93], it is worth reviewing here some of the technical possibilities.
· Object based etc. scale from slide presentation.
Results
Minor reservation
This work has shown many advantages of the OO approach for remote sensing applications, but there is one disadvantage. Current commercial OO GIS systems provide special easy-to-use languages for writing the methods, and these languages are interpreted. (Java is another example of an interpreted language.) This is not a problem for most purposes, but image analysis is particularly computation intensive and interpreted programs are always slower than compiled programs.
Thus in a practical system, once the algorithms have been prototyped in the special language ("LULL" in our case), they would need to be translated into C++ and linked in to the GIS software. This capability typically requires a slightly more expensive licence from the software vendor. However, making a commercial case for the expense of this translation and license should be easy as a fully-working image analysis system could be demonstrated.
Conclusions
Further Work
· Extension to more papers on Landscape Change would be possible (larger scale units: watersheds, agriculture v. urban, changes in statistical distribution of polygon geometry, farm sizes).
· Sandro is not so interested in time-series work, Philip thinks that handling historical/time data within the OO GIS (like the SsangYong land-parcel project) could be the basis for a further paper.
References
[Adam and Ganopadhyay 97] N.R. Adam and A. Ganopadhyay, Database Issues in Geographic Information Systems, Kluwer Academic Publ. (1997). ISBN 0-7923-9924-2
[Batty 93] Batty, P. "Object-Orientation - some objectivity please!". GIS 93 Conference Proceedings, May 1993, Birmingham, UK. http://www.smallworld.co.uk/
[Chance et al. 96] Arthur Chance, Richard G. Newell and David G. Theriault. An Overview of Smallworld Magik, Smallworld Technical Paper no.9. http://www.smallworld.co.uk/
[CORINE] CORINE Land Cover Technical Guide ISBN 92-826-2578-8
[Laser-Scan 94] Laser-Scan Ltd. "The Gothic Versioned Object-Oriented Database: an Introduction". November 1994, Laser-Scan, Cambridge UK. http://www.laser-scan.co.uk/papers
[Hartnall and MacAllister 95] Hartnall, T J and MacAllister, B. "Meeting Utility Needs Using Object-Oriented Spatial Technology". GIS 95 Conference Proceedings, May 1995, Birmingham, UK.http://www.laser-scan.co.uk/papers/utility.html
[Newell 92] Richard G. Newell. Practical experiences of using object-orientation to implement a GIS, Proceedings of GIS/LIS 92.
[Meyer 97] Meyer, Bertrand. Object-oriented Software Construction, 2nd Edition. Eaglewood Cliffs. Prentice Hall Co, 1997
[Muller 91] Muller, J-C. "Generalisation of Spatial Databases", in "Geographical Information Systems Volume 1: Principles" edited by David J Maguire, Michael F Goodchild and David W Rhind, pp457-75, Longmans, 1991.
[Perdigao and Annoni 97] V. Perdigao and A. Annoni, "Technical and Methodological Guide for Updating CORINE land COver Data Base", Publ. Joint Research Centre, European Environment Agency, EUR 17288 EN
[Smits and Dellepiane 97a] P.C. Smits and S.G. Dellepiane, "Synthetic Aperture Radar image segmentation by a detail preserving Markov Random Field approach," IEEE Transactions on Geoscience and Remote Sensing, Vol. 35, No. 4, July 1997, pp. 844-857.
[Smits and Dellepiane 97b] P.C. Smits and S. Dellepiane, "An Irregular MRF region label model for multi-channel image segmentation," Pattern Recognition Letters (18) 11-13 (1997), pp. 1133-1142.
[Smits 98] P.C. Smits, Model regularization in remote-sensing image analysis, Ph.D. thesis, University of Genoa, Italy (Genoa: Neos Edizioni Srl), 165 p., ISBN 88-87262-03-9. http://dibe.unige.it/tmr_smits
[Smits et al. 98] P.C. Smits, S.G. Dellepiane, and R. Schowengerdt, "Quality assessment of image classification algorithms for land-cover mapping: a review and a proposal for a cost-based approach," Int. Journal of Remote Sensing (accepted with minor revisions).
[Woodsford 96] Woodsford, P. A., 1996. Spatial Database Update - the Key to Effective Automation. International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4, Vienna 1996, pp 955-61.
[Woodsford and Hardy 97] Woodsford, P. A. and Hardy P.G. "Databases for Cartography and Navigation", International Cartographic Association conference proceedings, June 1997, Stockholm, Sweden.
Distributed Architectures
[Love 88] Love, K. J. Distributed Processing/Distributed Databases for GIS applications - Basic Concepts and Issues, Vol. 3, 1988 Annual Conference of the Urban and Regional Information Systems Association.
[Rhind and Green 88] Rhind, D. W. and Green, N. P. A. Design of a geographical information system for a heterogeneous scientific community, International Journal of Geographical Information Systems Vol. 2 No. 2.