Abstract:Hyperspectral image (HSI) classification methods based on deep learning (DL) techniques have succeeded wildly. However, the high-dimensional nonlinearity, spectral mixing, and difficulty in labeling training samples of HSI still hinder the accuracy of HSI classification. Several patch-free-based methods were proposed for HSI classification and have attracted attention. Nevertheless, recent patch-free methods have focused on low- or high-order interactions, ignoring the abundant middle-order interactions, failing to capture the multiorder interactions in the context of HSI, and difficulty extracting discriminative features when the sample data are imbalanced. In this article, a combining multiattention and asymmetric convolution (CMAAC) global learning framework was proposed for insufficient utilization of spectral–spatial information and imbalanced HSI samples. In CMAAC, the channel convolutional long short-term mechanism and multiorder spatial aggregation block (CLMS), which aim to capture multiorder interactions information and extract more discriminant spectral–spatial features effectively, is proposed. The pyramid-enhanced attention mechanism (PEAM) alleviates the information loss in feature flow, retains more spatial information, and better solves the challenge of scale diversity in different land-cover types. The asymmetric convolutional structure at the end of the model highlights the influence of local key feature points and improves its representation ability. Additionally, the framework introduces a joint loss function (JLF) to reduce misclassified classes caused by the imbalance between well-classified and hard-classified samples. We conducted experiments on four benchmark datasets and compared them with the state-of-the-art approaches, demonstrating the model's effectiveness and superiority.

What problem does this paper attempt to address?

The main problems that this paper attempts to solve are several key challenges in hyperspectral image (HSI) classification: 1. **High - dimensional non - linearity**: HSI has the characteristic of high - dimensionality, which leads to an increase in the non - linear complexity of data and makes the classification task more difficult. 2. **Spectral mixing**: The spectral information in HSI may become blurred due to the mixing of different substances, affecting the accuracy of classification. 3. **Difficulty in sample labeling**: The labeling of HSI requires professional knowledge and is time - consuming and labor - intensive, resulting in insufficient or unbalanced training samples. 4. **Insufficient utilization of multi - order interaction information**: Existing patch - free methods mainly focus on low - order or high - order interactions, ignoring the rich medium - order interaction information, resulting in the inability to comprehensively capture multi - order interaction information. 5. **Limited feature extraction ability**: In the case of unbalanced sample data, it is difficult to extract discriminative features. To address these challenges, the paper proposes a global learning framework (CMAAC) that combines multi - attention mechanisms and asymmetric convolution, aiming to more effectively utilize spectral - spatial information, solve the sample imbalance problem, and improve the accuracy of HSI classification. Specifically, the CMAAC framework contains the following core modules: - **Channel Convolution Long - Short - Term Mechanism and Multi - order Spatial Aggregation Block (CLMS)**: By combining the channel convolution long - short - term mechanism (CLSTM) and the multi - order spatial aggregation block (MSAB), CLMS can capture multi - order interaction information and extract more discriminative spectral - spatial features. - **Pyramid - enhanced Attention Mechanism (PEAM)**: PEAM expands the receptive field through parallel convolution and different dilation rates while retaining local detail features, effectively extracting fine - grained multi - scale spatial information and better solving the problem of scale diversity of different land cover types. - **Asymmetric Convolution Structure (ACS)**: ACS enhances the skeleton of the convolution kernel, restores the image edge information, and improves the representation ability of the model. - **Joint Loss Function (JLF)**: JLF is used to solve the class imbalance problem, reduce misclassified samples, and promote faster convergence of the model. Through these innovations, the experimental results of the CMAAC framework on four benchmark datasets show that it performs excellently in the HSI classification task and outperforms existing methods.

CMAAC: Combining Multiattention and Asymmetric Convolution Global Learning Framework for Hyperspectral Image Classification

AMFAN: Adaptive Multiscale Feature Attention Network for Hyperspectral Image Classification

A Cross-Channel Dense Connection and Multi-Scale Dual Aggregated Attention Network for Hyperspectral Image Classification

DMAF-NET: Deep Multi-Scale Attention Fusion Network for Hyperspectral Image Classification with Limited Samples

Pseudo Complex-Valued Deformable ConvLSTM Neural Network With Mutual Attention Learning for Hyperspectral Image Classification

CACFTNet: A Hybrid Cov-Attention and Cross-Layer Fusion Transformer Network for Hyperspectral Image Classification

Hyperspectral Image Classification based on Multi-Scale Convolutional Features and Multi-Attention Mechanisms

An Effective Hyperspectral Image Classification Network Based on Multi-Head Self-Attention and Spectral-Coordinate Attention

A Spectral-Spatial-Dependent Global Learning Framework for Insufficient and Imbalanced Hyperspectral Image Classification

Heterogeneous Spectral-Spatial Network with 3D Attention and MLP for Hyperspectral Image Classification Using Limited Training Samples

Multiscale Pixel-Level and Superpixel-Level Method for Hyperspectral Image Classification: Adaptive Attention and Parallel Multi-Hop Graph Convolution

Hyperspectral Image Classification With Multi-Attention Transformer and Adaptive Superpixel Segmentation-Based Active Learning

Hyperspectral Image Classification Framework Based on Multichannel Graph Convolutional Networks and Class-Guided Attention Mechanism

A Feature Embedding Network with Multiscale Attention for Hyperspectral Image Classification

Adaptive Multi-Feature Fusion Graph Convolutional Network for Hyperspectral Image Classification

Dual-Channel Convolution Network With Image-Based Global Learning Framework for Hyperspectral Image Classification

Hierarchical Multi-Scale Convolutional Neural Networks For Hyperspectral Image Classification

Cnn-assisted multi-hop graph attention network for hyperspectral image classification

A Dual Multi-Head Contextual Attention Network for Hyperspectral Image Classification

Hyperspectral Image Classification via Deep Network with Attention Mechanism and Multigroup Strategy

Hyperspectral Image Classification Based on 3D Coordination Attention Mechanism Network