Author/Editor		Laporterie, Florence; Flouzat, Guy; Amram, Olivier
Title		The morphological pyramid and its applications to remote sensing: multiresolution data analysis and features extraction
Type		članek
Source		Image Anal Stereol
Vol. and No.		Letnik 20, št. Suppl 1
Publication year		2001
Volume		str. 54-9
Language		eng
Abstract		ln remote sensing, image sources are more and more numerous, and spatial resolutions are higher and higher. Thus, stakes of the exploitation of those images increase, and a quick and accurate characterisation becomes indispensable. This paper deals with a method combining both theories of pyramid algorithm and mathematical morphology to study the physiographic characteristics of terrestrial ecosystems. A pyramid algorithm is processed thanks to morphological filters, then the extraction at each level of resolution of well-known landscapes features is performed. An application is shown on a digitised aerial photograph representing an heterogeneous landscape of orchards and forests along the Garonne river (France). Thanks to this example simulating very high spatial resolution imagery, the influence of the parameters of the pyramid can be, by definition, shown according to the spatial properties of the studied patterns. To conclude with, it can be seen that, when the morphological characteristics of landscape components are identified, the morphological pyramid approach is a promising attempt for multi-level features extraction by modelling geometrical relevant parameters.
Descriptors		ECOSYSTEM PHOTOGRAMMETRY IMAGE PROCESSING, COMPUTER-ASSISTED