Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2010, Vol. 7, No. 1, pp. 353-365
The algorithm and software suite for land cover types recognition based on locally-adaptive supervised classification
of satellite imagery
Space Research Institute of the Russian Academy of Sciences, 117997 Moscow, 84/32 Profsoyuznaya str
Efficient land cover types recognition using satellite imagery for the global and continental scale terrestrial ecosystems mapping strongly depends on availability of specialized classification techniques capable of taking objects' spectral characteristics spatial variability into account. The LAGMA supervised classification algorithm developed in RAS SRI meets the above requirements and is based on computing of local spectral signatures for various land cover types succeeded by the local-adaptive classification. The implementation of the method in a form of a specialized software suite provides the possibility of highly automated creation and update of land cover maps for extensive territories, such as large countries, continents or the whole world. The paper includes the functionality overview of the software suite and specifics of the software implementation and usage.
Keywords: satellite observation data, land cover types recognition, supervised classification, adaptive algorithms, distributed computing, software suite
Full text