HG-Mamba: A Hybrid Geometry-Aware Bidirectional Mamba Network for Hyperspectral Image Classification
Deep learning has demonstrated significant success in hyperspectral image (HSI) classification by effectively leveraging spatial–spectral feature learning. However, current approaches encounter three challenges: (1) high spectral redundancy and the presence of noisy bands, which impair the extractio...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-06-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/17/13/2234 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849427680754139136 |
|---|---|
| author | Xiaofei Yang Jiafeng Yang Lin Li Suihua Xue Haotian Shi Haojin Tang Xiaohui Huang |
| author_facet | Xiaofei Yang Jiafeng Yang Lin Li Suihua Xue Haotian Shi Haojin Tang Xiaohui Huang |
| author_sort | Xiaofei Yang |
| collection | DOAJ |
| description | Deep learning has demonstrated significant success in hyperspectral image (HSI) classification by effectively leveraging spatial–spectral feature learning. However, current approaches encounter three challenges: (1) high spectral redundancy and the presence of noisy bands, which impair the extraction of discriminative features; (2) limited spatial receptive fields inherent in convolutional operations; and (3) unidirectional context modeling that inadequately captures bidirectional dependencies in non-causal HSI data. To address these challenges, this paper proposes HG-Mamba, a novel hybrid geometry-aware bidirectional Mamba network for HSI classification. The proposed HG-Mamba synergistically integrates convolutional operations, geometry-aware filtering, and bidirectional state-space models (SSMs) to achieve robust spectral–spatial representation learning. The proposed framework comprises two stages. The first stage, termed spectral compression and discrimination enhancement, employs multi-scale spectral convolutions alongside a spectral bidirectional Mamba (SeBM) module to suppress redundant bands while modeling long-range spectral dependencies. The second stage, designated spatial structure perception and context modeling, incorporates a Gaussian Distance Decay (GDD) mechanism to adaptively reweight spatial neighbors based on geometric distances, coupled with a spatial bidirectional Mamba (SaBM) module for comprehensive global context modeling. The GDD mechanism facilitates boundary-aware feature extraction by prioritizing spatially proximate pixels, while the bidirectional SSMs mitigate unidirectional bias through parallel forward–backward state transitions. Extensiveexperiments on the Indian Pines, Houston2013, and WHU-Hi-LongKou datasets demonstrate the superior performance of HG-Mamba, achieving overall accuracies of 94.91%, 98.41%, and 98.67%, respectively. |
| format | Article |
| id | doaj-art-3d11b76b4aec4996b169054a98deb2ef |
| institution | Kabale University |
| issn | 2072-4292 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj-art-3d11b76b4aec4996b169054a98deb2ef2025-08-20T03:28:58ZengMDPI AGRemote Sensing2072-42922025-06-011713223410.3390/rs17132234HG-Mamba: A Hybrid Geometry-Aware Bidirectional Mamba Network for Hyperspectral Image ClassificationXiaofei Yang0Jiafeng Yang1Lin Li2Suihua Xue3Haotian Shi4Haojin Tang5Xiaohui Huang6School of Electronics and Communication Engineering, Guangzhou University, Guangzhou 510006, ChinaSchool of Electronics and Communication Engineering, Guangzhou University, Guangzhou 510006, ChinaSchool of Electronics and Communication Engineering, Guangzhou University, Guangzhou 510006, ChinaSchool of Electronics and Communication Engineering, Guangzhou University, Guangzhou 510006, ChinaSchool of Electronics and Communication Engineering, Guangzhou University, Guangzhou 510006, ChinaSchool of Electronics and Communication Engineering, Guangzhou University, Guangzhou 510006, ChinaSchool of Information Engineering, East China Jiaotong University, Nanchang 330044, ChinaDeep learning has demonstrated significant success in hyperspectral image (HSI) classification by effectively leveraging spatial–spectral feature learning. However, current approaches encounter three challenges: (1) high spectral redundancy and the presence of noisy bands, which impair the extraction of discriminative features; (2) limited spatial receptive fields inherent in convolutional operations; and (3) unidirectional context modeling that inadequately captures bidirectional dependencies in non-causal HSI data. To address these challenges, this paper proposes HG-Mamba, a novel hybrid geometry-aware bidirectional Mamba network for HSI classification. The proposed HG-Mamba synergistically integrates convolutional operations, geometry-aware filtering, and bidirectional state-space models (SSMs) to achieve robust spectral–spatial representation learning. The proposed framework comprises two stages. The first stage, termed spectral compression and discrimination enhancement, employs multi-scale spectral convolutions alongside a spectral bidirectional Mamba (SeBM) module to suppress redundant bands while modeling long-range spectral dependencies. The second stage, designated spatial structure perception and context modeling, incorporates a Gaussian Distance Decay (GDD) mechanism to adaptively reweight spatial neighbors based on geometric distances, coupled with a spatial bidirectional Mamba (SaBM) module for comprehensive global context modeling. The GDD mechanism facilitates boundary-aware feature extraction by prioritizing spatially proximate pixels, while the bidirectional SSMs mitigate unidirectional bias through parallel forward–backward state transitions. Extensiveexperiments on the Indian Pines, Houston2013, and WHU-Hi-LongKou datasets demonstrate the superior performance of HG-Mamba, achieving overall accuracies of 94.91%, 98.41%, and 98.67%, respectively.https://www.mdpi.com/2072-4292/17/13/2234hyperspectralimage classificationdeep learningMambageometry-aware |
| spellingShingle | Xiaofei Yang Jiafeng Yang Lin Li Suihua Xue Haotian Shi Haojin Tang Xiaohui Huang HG-Mamba: A Hybrid Geometry-Aware Bidirectional Mamba Network for Hyperspectral Image Classification Remote Sensing hyperspectralimage classification deep learning Mamba geometry-aware |
| title | HG-Mamba: A Hybrid Geometry-Aware Bidirectional Mamba Network for Hyperspectral Image Classification |
| title_full | HG-Mamba: A Hybrid Geometry-Aware Bidirectional Mamba Network for Hyperspectral Image Classification |
| title_fullStr | HG-Mamba: A Hybrid Geometry-Aware Bidirectional Mamba Network for Hyperspectral Image Classification |
| title_full_unstemmed | HG-Mamba: A Hybrid Geometry-Aware Bidirectional Mamba Network for Hyperspectral Image Classification |
| title_short | HG-Mamba: A Hybrid Geometry-Aware Bidirectional Mamba Network for Hyperspectral Image Classification |
| title_sort | hg mamba a hybrid geometry aware bidirectional mamba network for hyperspectral image classification |
| topic | hyperspectralimage classification deep learning Mamba geometry-aware |
| url | https://www.mdpi.com/2072-4292/17/13/2234 |
| work_keys_str_mv | AT xiaofeiyang hgmambaahybridgeometryawarebidirectionalmambanetworkforhyperspectralimageclassification AT jiafengyang hgmambaahybridgeometryawarebidirectionalmambanetworkforhyperspectralimageclassification AT linli hgmambaahybridgeometryawarebidirectionalmambanetworkforhyperspectralimageclassification AT suihuaxue hgmambaahybridgeometryawarebidirectionalmambanetworkforhyperspectralimageclassification AT haotianshi hgmambaahybridgeometryawarebidirectionalmambanetworkforhyperspectralimageclassification AT haojintang hgmambaahybridgeometryawarebidirectionalmambanetworkforhyperspectralimageclassification AT xiaohuihuang hgmambaahybridgeometryawarebidirectionalmambanetworkforhyperspectralimageclassification |