3DSS-Mamba: 3D-Spectral-Spatial Mamba for Hyperspectral Image Classification

TL;DR

This paper tackles hyperspectral image classification by addressing the inefficiencies of deep CNNs and Transformers in capturing global spectral-spatial dependencies. It introduces 3DSS-Mamba, a State Space Model–based framework that tokenizes 3D HSI data into spectral-spatial tokens (SSTG) and models interactions via a 3DSS mechanism with multiple scanning routes, implemented in stacked 3DMB blocks. The approach achieves state-of-the-art results on Pavia University, Indian Pines, and Houston 2013, while maintaining linear computational complexity relative to sequence length, improving scalability for high-dimensional data. The combination of global spectral-spatial modeling and efficiency has practical implications for robust, scalable HSI classification in remote sensing applications.

Abstract

Hyperspectral image (HSI) classification constitutes the fundamental research in remote sensing fields. Convolutional Neural Networks (CNNs) and Transformers have demonstrated impressive capability in capturing spectral-spatial contextual dependencies. However, these architectures suffer from limited receptive fields and quadratic computational complexity, respectively. Fortunately, recent Mamba architectures built upon the State Space Model integrate the advantages of long-range sequence modeling and linear computational efficiency, exhibiting substantial potential in low-dimensional scenarios. Motivated by this, we propose a novel 3D-Spectral-Spatial Mamba (3DSS-Mamba) framework for HSI classification, allowing for global spectral-spatial relationship modeling with greater computational efficiency. Technically, a spectral-spatial token generation (SSTG) module is designed to convert the HSI cube into a set of 3D spectral-spatial tokens. To overcome the limitations of traditional Mamba, which is confined to modeling causal sequences and inadaptable to high-dimensional scenarios, a 3D-Spectral-Spatial Selective Scanning (3DSS) mechanism is introduced, which performs pixel-wise selective scanning on 3D hyperspectral tokens along the spectral and spatial dimensions. Five scanning routes are constructed to investigate the impact of dimension prioritization. The 3DSS scanning mechanism combined with conventional mapping operations forms the 3D-spectral-spatial mamba block (3DMB), enabling the extraction of global spectral-spatial semantic representations. Experimental results and analysis demonstrate that the proposed method outperforms the state-of-the-art methods on HSI classification benchmarks.

Figures (12)

• Figure 1: The motivation of the proposed 3DSS-Mamba. Mamba modeling has demonstrated substantial potential in low-dimensional scenarios such as 1D language and 2D visual tasks, motivating its adaptability to high-dimensional HSI classification task.
• Figure 2: (a) The overall architecture of the proposed 3D-Spectral-Spatial Mamba (3DSS-Mamba) for HSI classification, which consists of a Spectral-Spatial Token Generation module (SSTG), ${N_L}$ stacked 3D-Spectral-Spatial Mamba Blocks (3DMB), and a classifier module; (b) The structural flow of proposed 3D-Spectral-Spatial Mamba Block (3DMB); (c) The computational procedure of proposed 3D-Spectral-Spatial Selective Scanning (3DSS).
• Figure 3: Five flattening routes are constructed to explore the impact of dimension prioritization.
• Figure 4: The detail structure of Spectral-Spatial Token Generation module (SSTG).
• Figure 5: Sensitivity analysis for the proposed method with different sizes of input patches.