Date of Publication:2017-01-01

Journal:海洋环境科学

Volume:36

Issue:1

Page Number:114-120

ISSN No.:1007-6336

Abstract:Mangrove species classification is important for studying the changes of
   mangrove ecosystem. In this paper, the distribution of mangroves in
   Tieshangang is chosen as the study area, and domestic ZY3 mapping
   satellite data is adopted as the data source. The spectral
   characterization of various mangroves is analyzed, four different
   vegetation indices (RVI,NDVI,VARI and NDGI) are exacted to add
   vegetation information, and the joint sparse representation algorithm is
   utilized to distinguish seven mangrove species. We analyze the
   vegetation index of seven mangroves (Aegiceras corniculatum,Excoecaria
   agallocha,Avicennia marina,Rhizophora stylosa,Kandelia candel,Sonneratia
   apetala and Bruguiear gymnorrhiza) and other species (bushwood,mudbank
   and grassland). The geometric dimension and spectral dimension are
   combined and the joint sparse representation is used for classification.
   The overall accuracy reached 95.37% and kappa coefficient reached 0.9347
   when we use the spectral data incorporate with four vegetation indices.
   Experiments show that spectral features combined with vegetation indices
   can improve classification accuracy, and NDVI has a greater contribution
   than other indices to distinguish mangrove species. Furthermore, joint
   sparse representation classifier has a good performance in mangrove
   species classification.

Note:新增回溯数据