MVNet：基于混合 Mamba-Transformer 视觉骨干架构的高光谱遥感图像分类
MVNet: Hyperspectral Remote Sensing Image Classification Based on Hybrid Mamba-Transformer Vision Backbone Architecture
AI总结：
本文提出 MVNet 架构，融合 3D-CNN、Transformer 和 Mamba 的优势，通过双分支 Mamba 模块与优化的 HSI-MambaVision Mixer 实现高效双向空间-光谱特征提取，在多个高光谱数据集上取得领先的分类精度与计算效率。
摘要：
高光谱图像（HSI）分类面临高维数据、训练样本有限和光谱冗余等挑战，这些问题往往导致过拟合和泛化能力不足。本文提出了一种新颖的 MVNet 网络架构，该架构融合了 3D-CNN 的局部特征提取能力、Transformer 的全局建模能力以及 Mamba 的线性复杂度序列建模能力，实现了高效的空间-光谱特征提取与融合。MVNet 设计了一个重构的双分支 Mamba 模块，包括一个状态空间模型（SSM）分支和一个采用 1D 卷积与 SiLU 激活的非 SSM 分支，增强了对短程和长程依赖关系的建模能力，同时降低了传统 Mamba 的计算延迟。优化后的 HSI-MambaVision Mixer 模块克服了因果卷积的单向限制，通过解耦注意力聚焦高价值特征，在单次前向传播中捕获双向空间-光谱依赖关系，缓解了参数冗余和维度灾难问题。在 IN、UP 和 KSC 数据集上，MVNet 在分类精度和计算效率方面均优于主流高光谱图像分类方法，展示了其处理复杂 HSI 数据的鲁棒能力。
英文摘要：
Hyperspectral image (HSI) classification faces challenges such as high-dimensional data, limited training samples, and spectral redundancy, which often lead to overfitting and insufficient generalization capability. This paper proposes a novel MVNet network architecture that integrates 3D-CNN's local feature extraction, Transformer's global modeling, and Mamba's linear complexity sequence modeling capabilities, achieving efficient spatial-spectral feature extraction and fusion. MVNet features a redesigned dual-branch Mamba module, including a State Space Model (SSM) branch and a non-SSM branch employing 1D convolution with SiLU activation, enhancing modeling of both short-range and long-range dependencies while reducing computational latency in traditional Mamba. The optimized HSI-MambaVision Mixer module overcomes the unidirectional limitation of causal convolution, capturing bidirectional spatial-spectral dependencies in a single forward pass through decoupled attention that focuses on high-value features, alleviating parameter redundancy and the curse of dimensionality. On IN, UP, and KSC datasets, MVNet outperforms mainstream hyperspectral image classification methods in both classification accuracy and computational efficiency, demonstrating robust capability in processing complex HSI data.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
