Construção de objetos espaciais Fuzzy a partir de imagens de satélite

Abstract

Fuzzy logic and theory have received increasing interest in Geographic Information Systems (GIS) environments. The representation of spatial data types using fuzzy set theory has led to the creation and definition of spatial fuzzy data types. Some implementations that enable the use of fuzzy spatial objects have been introduced, such as the fsr package, aiming to enable the use of fuzzy spatial algebras, such as the Spatial Plateau Algebra (SPA). In this context, there is already a path taken and consolidated for the construction of fuzzy spatial objects from point sets. However, there is still the challenge of building such types of objects from matrix spatial data, especially satellite images. Aiming to fill this gap, this work presents the inclusion of a fuzzification policy that allows to build SPA-based fuzzy spatial objects from matrix data. For this, a type of soft classification called Linear Spectral Mixture Model was used, which assigns membership degrees of each pixel to each of the mapping classes contained in an image. The result was incorporated into the fsr package in the spa creator function to generate fuzzy spatial objects and enable their use in GIS environments