Organizations: Department of Geomatics Engineering, University of Calgary, Canada · Department of Geography, Faculty of Arts, The University of Jordan, Amman, Jordan
Recently, DeepSeek has invented the manifold-constrained hyper-connection (mHC) approach which has demonstrated significant improvements over the traditional residual connection in deep learning models \cite{xie2026mhc}. Nevertheless, this approach has not been tailor-designed for improving hyperspectral image (HSI) classification. This paper presents a clustering-guided mHC Mamba model (mHC-HSI) for enhanced HSI classification, with the following contributions. First, to improve spatial-spectral feature learning, we design a novel clustering-guided Mamba module, based on the mHC framework, that explicitly learns both spatial and spectral information in HSI. Second, to decompose the complex and heterogeneous HSI into smaller clusters, we design a new implementation of the residual matrix in mHC, which can be treated as soft cluster membership maps, leading to improved explainability of the mHC approach. Third, to leverage the physical spectral knowledge, we divide the spectral bands into physically-meaningful groups and use them as the "parallel streams" in mHC, leading to a physically-meaningful approach with enhanced interpretability. The proposed approach is tested on benchmark datasets in comparison with the state-of-the-art methods, and the results suggest that the proposed model not only improves the accuracy but also enhances the model explainability. Code is available here: https://github.com/GSIL-UCalgary/mHC_HyperSpectral
Figures & tables
Fig. 1: Figure from [ 1 ] . Comparison between (a) a standard residual connection and (b) the Manifold-Constrained Hyper-Connections ( m HC) framework. The three red bounding boxes in (b) highlight the key components where our contributions are introduced. Specifically, they correspond to: (1) a clustering-guided Mamba module for enhanced spatial–spectral feature learning in hyperspectral images, (2) a novel residual mapping implementation that can be interpreted as soft cluster membership maps to decompose heterogeneous scenes and improve model explainability, and (3) a physically meaningful multi-stream design based on spectral band grouping to incorporate spectral prior knowledge. Together, these modifications adapt the original m HC framework for hyperspectral image classification with improved accuracy and interpretability.
Fig. 2: The architecture of the proposed clustering-guided m HC Mamba ( m HC-HSI) model. Based on the m HC framework, we implement F(⋅) with a novel clustering-guided Mamba module, which has a spectral Mamba module followed by the clustering-guided spatial Mamba module. Second, we implement the residual matrix Hres as soft cluster membership maps, which are used to decompose the complex and heterogeneous HSI into smaller clusters in the clustering-guided spatial Mamba. Third, we divide the input the spectral bands in the input HSI into physically-meaningful groups and use them as the ”streams” in xl .
Class No.
Train Number
Test Number
CNN-based
GAN-based
Transformer-based
Mamba-based
m HC-based
SSRN
SS-ConvNeXt
MTGAN
SSFTT
SSTN
GSC-ViT
MambaHSI
3DSS-Mamba
Ours
1
5
37
90.98
84.21
84.63
94.18
94.44
94.59
94.74
95.12
100.00
2
143
1142
97.54
99.53
97.20
94.9
97.99
95.19
98.08
93.22
99.13
3
83
664
96.88
96.62
95.9
92.16
95.48
96.54
96.36
93.44
98.24
4
24
189
97.7
95.24
96.85
94.8
93.12
99.47
90.05
96.71
100.00
5
48
387
95.13
97.91
95.61
95.33
99.22
98.45
97.67
96.71
95.82
TABLE I: Quantitative performance of different classification methods in terms of OA, AA, k , as well as the accuracies for each class on the Indian Pines dataset with 10 % training samples. The best results are in bold and colored shadow.
Fig. 3: The Indian Pines classification map generated by different methods. (a) SSRN (b) SS-ConvNeXt (c) MTGAN (d) SSFTT (e) SSTN (f) GSC-ViT (g) MambaHSI (h) 3DSS-Mamba (i) m HC-HSI (j) False Color Image (k) Ground Truth. Some red circles are shown on the RGB image to illustrate the boundary preservation of our proposed model.
Metrics
Expansion rate n =2
Expansion rate n =4
OA
98.95
98.54
AA
97.65
97.77
Kappa
95.98
95.80
TABLE II: Impact on the expansion rate n .
Fig. 4: Visualization of the interpretable Hres with overlaid class boundaries. The boundary is selected based on the highest mean value for each class boundary. White text shows the name of the category.
Although hyperspectral image (HSI) classification is critical for supporting various environmental applications, it is a challenging task due to the spectral-mixture effect, the spatial-spectral heterogeneity and the difficulty to preserve class boundaries and details. This letter presents a novel unmixing-guided spatial-spectral Mamba with clustering tokens for improved HSI classification, with the following contributions. First, to disentangle the spectral mixture effect in HSI for improved pattern discovery, we design a novel spectral unmixing network that not only automatically learns endmembers and abundance maps from HSI but also accounts for endmember variabilities. Second, to generate Mamba token sequences, based on the clusters defined by abundance maps, we design an efficient Top-K token selection strategy to adaptively sequence the tokens for improved representational capability. Third, to improve spatial-spectral feature learning and detail preservation, based on the Top-K token sequences, we design a novel unmixing-guided spatial-spectral Mamba module that greatly improves traditional Mamba models in terms of token learning and sequencing. Fourth, to learn simultaneously the endmember-abundance patterns and classification labels, a multi-task scheme is designed for model supervision, leading to a new unmixing-classification framework that outputs not only accurate classification maps but also a comprehensive spectral-library and abundance maps. Comparative experiments on four HSI datasets demonstrate that our model can greatly outperform the other state-of-the-art approaches. Code is available at https://github.com/GSIL-UCalgary/Unmixing_guided_Mamba.git
Yimin Zhu, Lincoln Linlin Xu
Department of Geomatics Engineering, University of Calgary, Canada
Although hyperspectral images (HSIs) provide rich spectral-spatial information, accurate pixel-level classification remains challenging because of spectral-spatial heterogeneity and complex spatial structures. Existing vision state space models (Mamba) typically construct sequences according to predefined spatial neighborhoods, without explicitly accounting for semantic similarity or spatial non-stationarity. To address this limitation, we propose Token Clustering and Semantic Sequence Mamba (STMamba), which organizes sparse tokens into semantically coherent sequences for hyperspectral image classification with the following features. First, at the macro level, a hierarchical encoder decoder progressively selects semantic tokens with the Token Clustering Module (TCM) and restores dense features using a parameter-free Cross-scale Neighborhood Attention (CNA) Upsampler. Second, at the micro level, TCM first identifies representative cluster centers through density-aware clustering and estimates soft memberships based on feature similarity. A quadtree-based dynamic selection strategy then retains sparse and spatially distributed tokens from each semantic cluster, forming coherent semantic-token sequences while reducing redundant pixel-wise representations. Third, parallel Spatial and Spectral Semantic-wise Sequencing Mamba (SWSM) modules capture complementary long-range spatial and spectral dependencies within homogeneous semantic token sequences while suppressing irrelevant interactions across heterogeneous regions. Experimental results on three large-scale benchmark datasets demonstrate that STMamba outperforms the SOTA methods with respect to quantitative and qualitative results.
Yimin Zhu, Mahmood Elahi, Lincoln Linlin Xu
Department of Geomatics Engineering, University of Calgary, Canada · Department of Electrical and Software Engineering, University of Calgary, Canada
Hyperspectral image (HSI) classification remains challenging due to high spectral dimensionality, redundancy, and limited labeled data. Although convolutional neural networks (CNNs) and Vision Transformers (ViTs) achieve strong performance by exploiting spectral-spatial information and long-range dependencies, they often incur high computational cost and large model size, limiting practical use. To address these limitations, a unified hybrid framework, termed ConvVitMamba, is proposed for efficient HSI classification. The architecture integrates three components: a multiscale convolutional feature extractor to capture local spectral, spatial, and joint patterns; a Vision Transformer based tokenization and encoding stage to model global contextual relationships; and a lightweight Mamba inspired gated sequence mixing module for efficient content-aware refinement without quadratic self-attention. Principal Component Analysis (PCA) is used as preprocessing to reduce redundancy and improve efficiency. Experiments on four benchmark datasets, including Houston and three UAV borne QUH datasets (Pingan, Qingyun, and Tangdaowan), demonstrate that ConvVitMamba consistently outperforms CNN, Transformer, and Mamba based methods while maintaining a favorable balance between accuracy, model size, and inference efficiency. Ablation studies confirm the complementary contributions of all components. The results indicate that the proposed framework provides an effective and efficient solution for HSI classification in diverse scenarios. The source code is publicly available at https://github.com/mqalkhatib/ConvVitMamba
Mohammed Q. Alkhatib
College of Engineering and IT, University of Dubai, Dubai 14143, United Arab Emirates