Classifying Pixels by Means of Fuzzy Relations

A Coloring Algorithm for Image Classification

“…Many different approaches can be found in remote sensing classification literature. In Amo et al 3,5 , for example, some of the authors proposed a classification model based upon a modified outranking model, basically taken from Pearman et al 12 . But the output information appeared to be difficult to be managed by non-qualified decision makers.…”

Fuzzy sets for image processing and understanding

“…These aspects have been largely developed in the literature (see e.g. [6,7,17,19,48,89,96,100,150]). Functional approaches consist in minimizing or maximizing a functional, which can be interpreted as an analytical representation of some objective.…”

A coloring fuzzy graph approach for image classification

“…The above algorithm has been applied, as a first experiment, to a standard image taken from the MATLAB package (a water colored tree, already considered by in [3], reproduced here as a grey picture (see Fig. 2)).…”

“…A key argument, at least in land cover problems (see [1,3,5]), is that our fuzzy classes have a core of connected pixels, surrounded by pixels showing some degree of membership to that core. Hence, classification is made pixel by pixel, but behavior of surrounding pixels should have a strong influence at every stage.…”

“…Many different approaches can be found in remote sensing classification literature. In [3,5], for example, the authors propose a classification model based upon a modified outranking model, basically taken from [21]. This approach requires decision makers to estimate quite a number of parameters, and therefore the learning process finally shown is unrealistic.…”