Konferenzbeitrag
Intelligent Image Database for Nature Managment Optimization
Vorschaubild nicht verfügbar
Volltext URI
Dokumententyp
Text/Conference Paper
Zusatzinformation
Datum
2000
Autor:innen
Zeitschriftentitel
ISSN der Zeitschrift
Bandtitel
Verlag
Metropolis
Zusammenfassung
Intelligent image databases have to facilitate analyses of space and aerial observations of the environment processes and phenomena for estimation the environment current state. Such image databases should efficiently combine the experts' image interpretations and the corresponding environment knowledge. The paper demonstrates a new application of computer vision to image databases the use of image texture for annotation, the description of content. The goal was to use a scheme that is able to automatically decide on the image features and based upon psychophysical studies of human perception nature and computer vision models in contrast to multiple cue-based schemes being still heuristic. The approach provides a learning algorithm for selecting the most representative features of the homogeneous and piecewise-homogeneous data. Highly specializes and context-dependent features are extracted automatically and spatial information is preserved.