DER

Título:

DER :  dynamic evidential reasoning applied to hyperspectral images classification

Autor:

Sanz, Cecilia Verónica

Colaboradores:

; Jordan, Ramiro

Temas:

IMÁGENESRECONOCIMIENTO Y CLASIFICACIÓN

En:

Journal of Computer Science & Technology, 1(6)

Resumen:

This paper describes a new classification method (DER) based on evidential reasoning to which a series of modifications are added [1]. DER allows including new evidence for the classification process and defines a different decision rule. The evidential reasoning algorithm provides a means to combine evidence from different data sources. It is a supervised classification technique that uses a training samples set. This novel method (DER) offers a learning stage to introduce new evidence in case the classifier requires so. Moreover, it uses the plausibility measure in order to define the decision rule as a way to incorporate data- associated uncertainty. The proposed method is applied in order to classify crops in hyperspectral images of the area of Nebraska (USA). Some results obtained are presented in order to assess DER precision.

URL/DOI:

http://goo.gl/1RJ32j

Medio:

Soporte electrónico

Tipo de documento:

Artículo

Descripción física:

1 archivo (67 KB)

Idioma:

Inglés

Publicación:

Universidad Nacional de La Plata. Facultad de Informática, 2002

Puede solicitar más fácilmente el ejemplar con: A0595

Ver estantes