Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

HSLiNets: Evaluating Band Ordering Strategies in Hyperspectral and LiDAR Fusion

Judy X Yang, Jing Wang, Zhuanfeng, Li, Chenhong Sui Zekun Long, Jun Zhou

TL;DR

The paper tackles the unexplored influence of spectral band order on hyperspectral-LiDAR fusion for land-cover classification. It introduces HSLiNets, a dual-branch architecture that processes HSI in ascending and descending band orders and fuses the resulting features with LiDAR through a shared classifier, enabling learning from multiple spectral sequences. The study defines four band-order configurations and a dual-pair fusion strategy to maximize discriminative information, achieving state-of-the-art OA and AA on Houston 2013 and Trento, with robustness across architectures and patch sizes. The work highlights band-order as a critical design factor in multi-modal remote sensing and provides data and code for reproducibility.

Abstract

The integration of hyperspectral imaging (HSI) and Light Detection and Ranging (LiDAR) data provides complementary spectral and spatial information for remote sensing applications. While previous studies have explored the role of band selection and grouping in HSI classification, little attention has been given to how the spectral sequence or band order affects classification outcomes when fused with LiDAR. In this work, we systematically investigate the influence of band order on HSI-LiDAR fusion performance. Through extensive experiments, we demonstrate that band order significantly impacts classification accuracy, revealing a previously overlooked factor in fusion-based models. Motivated by this observation, we propose a novel fusion architecture that not only integrates HSI and LiDAR data but also learns from multiple band order configurations. The proposed method enhances feature representation by adaptively fusing different spectral sequences, leading to improved classification accuracy. Experimental results on the Houston 2013 and Trento datasets show that our approach outperforms state-of-the-art fusion models. Data and code are available at https://github.com/Judyxyang/HSLiNets.

HSLiNets: Evaluating Band Ordering Strategies in Hyperspectral and LiDAR Fusion

TL;DR

The paper tackles the unexplored influence of spectral band order on hyperspectral-LiDAR fusion for land-cover classification. It introduces HSLiNets, a dual-branch architecture that processes HSI in ascending and descending band orders and fuses the resulting features with LiDAR through a shared classifier, enabling learning from multiple spectral sequences. The study defines four band-order configurations and a dual-pair fusion strategy to maximize discriminative information, achieving state-of-the-art OA and AA on Houston 2013 and Trento, with robustness across architectures and patch sizes. The work highlights band-order as a critical design factor in multi-modal remote sensing and provides data and code for reproducibility.

Abstract

The integration of hyperspectral imaging (HSI) and Light Detection and Ranging (LiDAR) data provides complementary spectral and spatial information for remote sensing applications. While previous studies have explored the role of band selection and grouping in HSI classification, little attention has been given to how the spectral sequence or band order affects classification outcomes when fused with LiDAR. In this work, we systematically investigate the influence of band order on HSI-LiDAR fusion performance. Through extensive experiments, we demonstrate that band order significantly impacts classification accuracy, revealing a previously overlooked factor in fusion-based models. Motivated by this observation, we propose a novel fusion architecture that not only integrates HSI and LiDAR data but also learns from multiple band order configurations. The proposed method enhances feature representation by adaptively fusing different spectral sequences, leading to improved classification accuracy. Experimental results on the Houston 2013 and Trento datasets show that our approach outperforms state-of-the-art fusion models. Data and code are available at https://github.com/Judyxyang/HSLiNets.

Paper Structure

This paper contains 12 sections, 3 equations, 2 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Proposed Method Overview and Architecture Description
  4. Band Order Configurations
  5. Classification and Loss Function
  6. Experiments
  7. Experimental Setup
  8. Impact of Band Order on Classification
  9. Impact of HSI Band Order on Cross-Architecture Validation
  10. Comparison with State-of-the-Art Fusion Methods
  11. Impact of Patch Size on Classification Performance
  12. Conclusion

Figures (2)

  • Figure 1: Architectural overview of the proposed HSI-LiDAR fusion model. The framework consists of four main components: (A) Input patches( (HSI: $p \times p \times C$ and LiDAR: $p \times p \times 1$ are processed through dual streams. (B) B) Linear projection layers ($\mathbb{R}^C \rightarrow \mathbb{R}^{d_h}$) align modalities.(C) CNN blocks extract spectral-spatial features with band-order-specific parameters. (D) The fusion block combines streams before final classification.
  • Figure 2: OA and Patch Size Relationship based on Houston 2013 and Trento Datasets