Spectral fuzzy classification: a supervised approach

The goal of this paper is to present an algorithm for pattern recognition, leveraging on an existing fuzzy clustering algorithm developed by Del Amo et al. [3, 5], and modifying it to its supervised version, in order to apply the algorithm to different pattern recognition applications in Remote Sensing. The main goal is to recognize the ob ject and stop the search depending on the precision of the application. The referred algorithm was the core of a classification system based on Fuzzy Sets Theory (see [14]), approaching remotely sensed classification problems as multicriteria decision making problems, solved by means of an outranking methodology (see [12] and also [11]). The referred algorithm was a unsupervised classification algorithm, but now in this paper will present a modification of the original algorithm into a supervised version.