Resumo

KHELIFA, Said  e  BOUKHATEM, Fatima. Simulated Annealing-Based Optimization for Band Selection in Hyperspectral Image Classification. Comp. y Sist. [online]. 2023, vol.27, n.4, pp.873-879.  Epub 17-Maio-2024. ISSN 2007-9737.  https://doi.org/10.13053/cys-27-4-4519.

In this paper, a new optimization based framework for hyperspectral image classification problem is proposed. Band selection is a primordial step in supervised/unsupervised hyperspectral image classification. It attempts to select an optimal subset of spectral bands from the entire set of hyperspectral cube. This subset is considered as the relevant informative subset of bands. The advantage of an efficient band selection approach is to reduce the hughes phenomenon by removing irrelevant and redundant bands. In this study, we propose a new objective function for the band selection problem by using Simulated Annealing as an optimization method. The proposed approach is tested on three Hyperspectral Images largely used in the literature. Experimental results show the performance and efficiency of the proposed approach.

Palavras-chave : Optimization; band selection; classification; bagging; correlation; simulated annealing.

        · texto em Inglês     · Inglês ( pdf )