EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Leveraging potential of limpid attention transformer with dynamic tokenization for hyperspectral image classification

Dhirendra Prasad Yadav, Deepak Kumar, Anand Singh Jalal, Bhisham Sharma iD, Panos Liatsis iD

PLOS One · 2025

DOI: 10.1371/journal.pone.0328160 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, 2025. Verfasst von Dhirendra Prasad Yadav und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Computer Engineering & Applications, G.L.A. University, Mathurar, Uttar Pradesh, India. Hauptthemen laut MEDLINE: Neural Networks, Computer, Hyperspectral Imaging, Image Processing, Computer-Assisted. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Hyperspectral data consists of continuous narrow spectral bands. Due to this, it has less spatial and high spectral information. Convolutional neural networks (CNNs) emerge as a highly contextual information model for remote sensing applications. Unfortunately, CNNs have constraints in their underlying network architecture in regards to the global correlation of spatial and spectral features, making them less reliable for mining and representing the sequential properties of spectral signatures. In this article, limpid size attention network (LSANet) is proposed, which contains 3D and 2D convolution blocks for enhancement of spatial-spectral features of the hyperspectral image (HSI). In addition, limpid attention block (LAB) is designed to provide a global correlation of the spectral and spatial features through LS attention. Furthermore, the computational costs of LS-attention are less compared to the multi-head self-attention (MHSA) of the classical vision transformer (ViT). In the ViT encoder a conditional position encoding (CPE) module is utilized that dynamically generates tokens from the feature maps to capture a richer contextual representation. The LSANet obtained overall accuracy (OA) of 98.78%, 98.67%, 97.52% and 89.45%, respectively, on the Indian Pines (IP), Pavia University (PU), Salina Valley (SV) and Botswana datasets. Our model’s quantitative and qualitative results are considerably better than the classical CNN and transformer-based methods.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Neural Networks, ComputerHyperspectral ImagingImage Processing, Computer-AssistedAlgorithmsRemote Sensing Technology

Autor:innen und Institutionen

  • Department of Computer Engineering & Applications, G.L.A. University, Mathurar, Uttar Pradesh, India.
  • Department of Computer Engineering, NIT Meghalaya, Shillong, Meghalaya, India.
  • School of Computer Science & Information Technology, Devi Ahilya Vishwavidyalaya (DAVV), Indore, Northern Mariana Island, India.
  • Centre for Research Impact & Outcome, Chitkara University Institute of Engineering and Technology, Chitkara University, Rajpura, Punjab, India.
  • Center for Cyberphysical Systems, Department of Computer Science, Khalifa University, United Arab Emirates.

ORCID-Profile: Bhisham Sharma, Panos Liatsis

Zitieren

APA

Yadav, D. P., Kumar, D., Jalal, A. S., Sharma, B., & Liatsis, P. (2025). Leveraging potential of limpid attention transformer with dynamic tokenization for hyperspectral image classification. PLOS One. https://doi.org/10.1371/journal.pone.0328160

Vancouver

Yadav DP, Kumar D, Jalal AS, Sharma B, Liatsis P. Leveraging potential of limpid attention transformer with dynamic tokenization for hyperspectral image classification. PLOS One. 2025. doi:10.1371/journal.pone.0328160

BibTeX
@article{Yadav2025,
  author = {Yadav, Dhirendra Prasad and Kumar, Deepak and Jalal, Anand Singh and Sharma, Bhisham and Liatsis, Panos},
  title = {Leveraging potential of limpid attention transformer with dynamic tokenization for hyperspectral image classification},
  journal = {PLOS One},
  year = {2025},
  doi = {10.1371/journal.pone.0328160}
}
Kostenlos starten →