Uso de la teoría de lógica difusa en la clasificación de imágenes satelitales con coberturas mixtas: un caso de área urbana, Mérida, Venezuela
Cuando hacemos clasificación
supervisada de imágenes satelitales,
con el fin
de generar mapas de uso y cobertura, se nos presentan este tipo de problema en
áreas urbanas, agrícolas y ecotonos. La clasificación supervisada se ha hecho tradicionalmente con clasificadores “duros”, estos solamente nos permiten “entrenar” al sistema de
manera tal que un
determinado uso o cobertura es
por completo de determinada clase, sea bosque, urbano, agrícola, etc. lo
que en realidad no es enteramente cierto. Existen zonas de transición o usos mezclados en la superficie
terrestre. Se ha propuesto el uso de
En este trabajo se desarrollan dos tipos de clasificación
supervisada en imágenes satelitales en un área urbana, específicamente en
Palabras claves: Clasificación supervisada,
Imágenes satelitales,
Lógica Difusa, Mérida, Venezuela.
Use of theory of
Fuzzy Logic in the classification of satellite images with mixed covers: a case
of urban area,
The Theory of Fuzzy Logic is a
technique that has application in different fields of science. Describes
properties that have a continuous variation in their values, associating parts
from those values to a semantic label, for example, in the change that exists
when passing from a dense forest to a disperse forest doesn't pass in an
immediate way, or crisp limit, a transition area exists that cannot be
classified neither with a label neither with the other one, that area possesses
the characteristics of both forest types. In this case the Fuzzy Logic theory
is good to assign percentages of ownership, that is to say, maybe exist in that area 60% of dense forest and 40% of disperse
forest. Equal happens in agricultural and in urban areas.
When we make supervised
classification of satellite images, with the purpose of generating use and
covering maps, are
presented this type of problem in urban, agricultural areas and "ecotonos". The supervised classification has been made
traditionally with "hard"
classifiers, these only allow us "to train" the system such a way
that certain use or covering is completely of certain class, be forest, urban,
agricultural, etc., which in fact is not entirely true. Transition areas or mixed uses exists in the terrestrial surface. It is proposed
the use of the Theory of Fuzzy Logic to cope this problem.
In this research two
supervised classification types are developed with satellite images in an urban
area, specifically in the City of
Keywords: supervised
classification, satellite images, Fuzzy Logic,