Noisy Multispectral Image Classification using Feature Fusion

Alexandre L. M. LevadaNelson D. A. Mascarenhas

Methods for material analysis and identification on images are essential in many applications. In this paper we propose a method to improve unsupervised feature extraction on multispectral images with applications on soil science, combining second and higher order statistical methods. These images were obtained with transmission of different energies with a computerized tomograph scanner and the use of the linear attenuation coefficients as features for classification is studied. The results were evaluated by the leave-one-out cross-validation error and the Kappa coefficient from the confusion matrix.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: