%0 Conference Proceedings
%4 sid.inpe.br/mtc-m18/2012/05.16.17.10
%2 sid.inpe.br/mtc-m18/2012/05.16.17.10.27
%@isbn 978-85-17-00059-1
%T Classification of unmanaged forest reserves in Flanders (Belgium) at the tree crown level using airborne hyperspectral and LiDAR data
%D 2012
%A Devriendt, Flore,
%A Coillie, Frieke Van,
%A Wulf, Robert De,
%A Bertels, L.,
%A Kempeneers, P.,
%A Vandekerkhove, K.,
%A Morsdorf, F.,
%@electronicmailaddress flore.devriendt@ugent.be
%@electronicmailaddress frieke.vancoillie@ugent.be
%@electronicmailaddress robert.dewulf@ugent.be
%E Feitosa, Raul Queiroz,
%E Costa, Gilson Alexandre Ostwald Pedro da,
%E Almeida, Cláudia Maria de,
%E Fonseca, Leila Maria Garcia,
%E Kux, Hermann Johann Heinrich,
%B International Conference on Geographic Object-Based Image Analysis, 4 (GEOBIA).
%C Rio de Janeiro
%8 May 7-9, 2012
%I Instituto Nacional de Pesquisas Espaciais (INPE)
%J São José dos Campos
%P 141-145
%S Proceedings
%1 Instituto Nacional de Pesquisas Espaciais (INPE)
%K Hyperspectral, LiDAR, Image segmentation, Crown delineation, Classification, Unmanaged forests.
%X Focusing on unmanaged forest reserves in Flanders (Belgium), we present an approach for the generation of tree species maps at the tree crown level using hyperspectral CASI imagery and full waveform LiDAR data. Unmanaged forest reserves are characterized by growth stage diversity, high crown closure, multi-layering of the canopy and the non-existence of a pre-ordered spatial tree distribution. To be able to classify trees at the tree crown level in these highly structured closed forests, an algorithm for delineating tree crowns need to be developed that can handle a) expansive crowns with large branches exhibiting a huge amount of within-crown shadow, and b) crowns belonging to the sub-canopy that are partially overtopped or shadowed by neighbours. Using the LiDAR-derived canopy height model (CHM), a tree detection algorithm based on a Directional Local Filtering (DLF) is applied prior to crown delineation to automatically locate individual tree crowns. Tree crowns are subsequently outlined using object-based decision rules developed in eCognition. Following tree crown delineation, the mean-lit spectra derived from the hyperspectral CASI data of individual tree crowns are extracted to identify tree species.
%9 Forest Analysis
%@language en
%3 043.pdf