Oct 2026· 3rd International Conference on Electronic Materials and Information Engineering (EMIE 2023)· Vol 14382, pp. 1438214 - 1438214-5· 0 citations· 8 references
Engineering
TL;DR
RS-HCFormer is proposed, a hierarchical conditional encoding Transformer for hyperspectral image classification that substantially outperforms state-of-the-art methods in both accuracy and efficiency.
Abstract
Hyperspectral remote sensing images can characterize land cover attributes with high spectral resolution. However, their high dimensionality, spectral variability, and limited labeled samples pose significant challenges to accurate classification. Existing Transformer-based methods suffer from two major limitations: fixed position encodings cannot adapt to complex spatial structures, and the computational complexity of standard self-attention grows quadratically with input size. To address these issues, this paper proposes RS-HCFormer, a hierarchical conditional encoding Transformer for hyperspectral image classification. The main contributions are twofold. First, a conditional position encoding mechanism dynamically generates positional information based on local neighborhoods, enhancing adaptability to spatial structures. Second, a hierarchical window attention mechanism introduces learnable carrier tokens for each local window, enabling efficient local-global context interaction while reducing computational cost. RS-HCFormer employs a cascaded 3D-CNN and 2D-CNN to extract spatial-spectral features, followed by hierarchical conditional attention encoding for pixel-wise classification. Experimental results on multiple public datasets demonstrate that RS-HCFormer substantially outperforms state-of-the-art methods in both accuracy and efficiency.
This work proposes a four–stage hybrid framework for hyperspectral image classification that integrates spectral modeling and spatial feature extraction within a unified pipeline and compares the model with recent baselines, including DaViT.
The results suggest that decoupled spatial–spectral representation and adaptive multi-scale fusion can improve hyperspectral image classification under the evaluated benchmark settings.
Hyperspectral image classification is an important research topic in the field of remote sensing. Its core objective is to accurately identify different land-cover categories by jointly exploiting the spatial information and spectral characteristics of pixels. However, the complex and diverse spatial structures of grou...
Li-Guo Wang, Teng Wen, Dan-Feng Liu· Remote Sensing· 0 citations
A Position-Aware Attention Module (PAM) that employs MLP-based positional encoding to inject spatial structure information, thereby strengthening positional representation and a Query Interaction Mechanism (QIM) to enable complementary fusion of cross-domain features, where outputs of spatial and spectral branches serv...
Hai-Jun Wang, Hao-Yu Hu, Ya-Lin Nie et al.· International journal of pat...· 0 citations
A new method for HSI classification by integrating spectral band selection with advanced deep learning techniques is presented, and an optimized band selection method alongside data augmentation using crop-and-resize transformations to enhance feature diversity is proposed.
Eman N. Abdelhafez, Ahmed Hagag, Tamer A. Abassy· Scientific Reports· 0 citations
A novel hierarchical bilinear scattering Transformer (HBS-Former) that jointly captures local frequency-domain details and global semantic context and a hierarchical bilinear fusion (HBF) mechanism is developed to explicitly model second-order interactions among hierarchical features.
Hui-Hui Dong, Tong Wang, Zong-Fang Ma et al.· IEEE Transactions on Geoscie...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.