In this work, we evaluate different classification algorithms used for multi-target detection in hyperspectral imaging. We took into consideration the scenario of landmine detection in which we compared the performance of each method in various cases. In addition, we introduced the detection of targets using artificial intelligence-based methods in order to obtain better detection performance together with target identification and estimation of its abundance. These algorithms were tested on various types of hyperspectral images where the spectra of the landmines were planted in different proportions in the hyperspectral scenes. The results show the advantage of using our training strategy for radial basis function neural networks (RBFNN) in order to detect, identify and estimate the abundance of the targets in hyperspectral images at the same time. Moreover, the proposed technique requires a comparable computational cost with respect to state of art target detection techniques.

RBF Neural Network for Landmine Detection in Hyperspectral Imaging / Makki, Ihab; Younes, Rafic; Khodor, Mahdi; Khoder, Jihan; Francis, Clovis; Bianchi, Tiziano; Rizk, Patrick; Zucchetti, Massimo. - ELETTRONICO. - (2018), pp. 1-6. (Intervento presentato al convegno 2018 7th European Workshop on Visual Information Processing (EUVIP) tenutosi a Tampere, Finland nel 26-28 Nov. 2018) [10.1109/EUVIP.2018.8611652].

RBF Neural Network for Landmine Detection in Hyperspectral Imaging

Makki, Ihab;Khodor, Mahdi;Bianchi, Tiziano;Zucchetti, Massimo
2018

Abstract

In this work, we evaluate different classification algorithms used for multi-target detection in hyperspectral imaging. We took into consideration the scenario of landmine detection in which we compared the performance of each method in various cases. In addition, we introduced the detection of targets using artificial intelligence-based methods in order to obtain better detection performance together with target identification and estimation of its abundance. These algorithms were tested on various types of hyperspectral images where the spectra of the landmines were planted in different proportions in the hyperspectral scenes. The results show the advantage of using our training strategy for radial basis function neural networks (RBFNN) in order to detect, identify and estimate the abundance of the targets in hyperspectral images at the same time. Moreover, the proposed technique requires a comparable computational cost with respect to state of art target detection techniques.
2018
978-1-5386-6897-9
File in questo prodotto:
File Dimensione Formato  
makki_EUVIP2018_OA.pdf

accesso aperto

Descrizione: articolo versione autore
Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 480.89 kB
Formato Adobe PDF
480.89 kB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2724277
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo