Avtor/Urednik		Debeir, Olivier; Latinne, Patrice; van den Steen, Isabelle
Naslov		Remote sensing classification of spectral, spatial and contextual data using multiple classifier systems
Tip		članek
Vir		Image Anal Stereol
Vol. in št.		Letnik 20, št. Suppl 1
Leto izdaje		2001
Obseg		str. 584-9
Jezik		eng
Abstrakt		This study promotes the use of a multiclassifier system (MCS) fed with high-resolution remote sensing data coupled with contextual and textural data in the domain of land cover and land use classification. The gain of this approach is shown by a favorable comparison of our BAGFS classifier (a mixture of bagging and feature subset classifier) over two single classifier techniques (5-NN and C4.5 decision tree). The map chosen for the comparison is extracted from LANDSAT Tliematic Mapper data restricted on a Belgian area with a varied landscape and a 11 land cover legend.
Deskriptorji		SPACECRAFT URBANIZATION IMAGE PROCESSING, COMPUTER-ASSISTED DECISION TREES