Spectral-Spatial Features Exploitation Using Lightweight HResNeXt Model for Hyperspectral Image Classification

Hyperspectral image classification is vital for various remote sensing applications; however, it remains challenging due to the complex and high-dimensional nature of hyperspectral data. This paper introduces a novel approach to address this challenge by leveraging spectral and spatial features through a lightweight HResNeXt model. The proposed model is designed to overcome the limitations of traditional methods by combining residual connections and cardinality to enable efficient and effective feature extraction from hyperspectral images, capturing both spectral and spatial information simultaneously. Furthermore, the paper includes an in-depth analysis of the learned spectral-spatial features, providing valuable insights into the discriminative power of the proposed approach. The extracted features exhibit strong discriminative capabilities, enabling accurate classification even in challenging scenarios with limited training samples and complex spectral variations. Extensive experimental evaluations are conducted on four benchmark hyperspectral data sets, the Pavia university (PU), Kennedy Space Center (KSC), Salinas scene (SA), and Indian Pines (IP). The performance of the proposed method is compared with the state-of-the-art methods. The quantitative and visual results demonstrate the proposed approach's high classification accuracy, noise robustness, and computational efficiency superiority. The HResNeXt obtained an overall accuracy on PU, KSC, SA, and IP, 99.46%, 81.46%, 99.75%, and 98.64%, respectively. Notably, the lightweight HResNeXt model achieves competitive results while requiring fewer computational resources, making it well-suited for real-time applications. La classification d'images hyperspectrales est vitale pour diverses applications de teledetection. Cependant, cela reste difficile en raison de la nature complexe et de la haute dimensionnalite des donnees hyperspectrales. Cet article presente une nouvelle approche pour relever ce defi en tirant parti des caracteristiques spectrales et spatiales grace a un modele HResNeXt leger. Le modele propose est concu pour surmonter les limites des methodes traditionnelles en combinant les connexions residuelles et la cardinalite pour permettre une extraction efficace des caracteristiques des images hyperspectrales, capturant simultanement les informations spectrales et spatiales. En outre, l'article comprend une analyse approfondie des caracteristiques spectrales et spatiales apprises, fournissant des informations precieuses sur le pouvoir discriminatif de l'approche proposee. Les caracteristiques extraites presentent de fortes capacites discriminantes, permettant une classification precise meme dans des scenarios difficiles avec de petits echantillons d'entrainement et des variations spectrales complexes. Les evaluations experimentales ont ete menees sur quatre ensembles de donnees hyperspectrales de reference: PU, KSC, SA et IP. La performance de la methode proposee est comparee aux methodes de pointe. Les resultats quantitatifs et visuels demontrent une haute precision des classifications pour l'appropre proposee, sa robustesse au bruit et sa superiorite dans l'efficacite de calcul. Le HResNeXt a obtenu un OA sur PU, KSC, SA et IP, de 99,46%, 81,46%, 99,75% et 98,64%, respectivement. Notamment, le modele HResNeXt leger permet d'obtenir des resultats competitifs tout en necessitant moins de ressources de calcul, ce qui le rend bien adapte aux applications en temps reel.