177
The SVM algorithm aims to determine decision thresholds that produce optimum separation between classes by minimizing errors (Vapnik 1995). SVM is a
computational technique for pattern recognition problems, based on the optimal
separation of classes (Huang et al. 2002). Further details of the SVM method, as
well as applications in remote sensing area, may be found in Waske et al. (2010),
Liesenberg and Gloaguen (2013), and Garofalo et al. (2015).
5.4.1 Discussion
As a case study of a tropical coastal wetland it presents the following discussion of
the results on the use land and occupancy in Whale Coast region, later followed by
a proposition of possible diagnoses and scenarios.
5.4.2 Land Use and Occupation in Whale Coast, Bahia
Land use and occupation between 1984 and 2014 were observed by analyzing the
spectral behavior of the targets and the fieldwork performed in Whale Coast.
Figures 5.14 and 5.15 show the spatialized results, based on the categories previously described.
Fig. 5.14 Map on the use and occupation in the Coastal Plain of Caravelas in 1984
5 Land Use and Occupation of Coastal Tropical Wetlands: Whale Coast, Bahia, Brazil
Précédent

- 192/878

Suivant