Article
Computer Science, Interdisciplinary Applications
Linfeng Wu, Huajun Wang, Tong Zhang
Summary: In this paper, a multiscale 3D convolution with context attention network is proposed for HSI classification. The method introduces convolution kernels of different sizes to enlarge the receptive field and adaptively detect HSI features in different scales. Two subnetworks are built to efficiently exploit hierarchical spectral and spatial features and enhance feature transmission. Experimental results show that the proposed method outperforms state-of-the-art models on multiple benchmark HSI datasets in terms of overall accuracy.
EARTH SCIENCE INFORMATICS
(2022)
Article
Environmental Sciences
Yi Liu, Jian Zhu, Jiajie Feng, Caihong Mu
Summary: This paper proposes a hyperspectral image classification method based on a Feature Embedding Network with Multiscale Attention. By designing a Multiscale Attention Module and introducing an Adaptive Spatial Feature Fusion strategy, the method is able to extract and fuse features at different depths, resulting in better classification accuracies compared to other methods.
Article
Agriculture, Multidisciplinary
Ying Meng, Zheng Ma, Zeguang Ji, Rui Gao, Zhongbin Su
Summary: This paper obtains canopy hyperspectral images of rice varieties using a UAV platform with a S185 hyperspectral imaging device. A hybrid convolutional neural network structure is used to automatically analyze and extract spectral and spatial features of 14 rice varieties. Additionally, an attention module is applied to optimize the model. Extensive experiments demonstrate the validity of the proposed model, achieving high accuracy in fine classification of rice varieties. The model contributes to automatic field identification and crop phenotype research, and presents new possibilities for the development of precision agriculture and smart agriculture.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Computer Science, Artificial Intelligence
Jie Feng, Zhanwei Ye, Shuai Liu, Xiangrong Zhang, Jiantong Chen, Ronghua Shang, Licheng Jiao
Summary: Band selection is a crucial task in hyperspectral image processing. The proposed dual-graph convolutional network utilizes band attention and sparse constraint to effectively select bands, extract features, and perform classification. Experimental results demonstrate the superiority of this method over current state-of-the-art band selection methods.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Remote Sensing
Ting Tang, Jiangping Liu, Xiaoling Luo, Xiaojing Gao, Xin Pan
Summary: Recently, the classification of hyperspectral images has been extensively studied. This paper proposes a triple-branch ternary-attention mechanism network with deformable 3D convolution to effectively utilize the spectral and spatial information. Experimental results demonstrate that the proposed method outperforms existing algorithms on different datasets.
INTERNATIONAL JOURNAL OF REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Linzhou Yu, Jiangtao Peng, Na Chen, Weiwei Sun, Qian Du
Summary: This paper proposes a novel two-branch deeper graph convolutional network (TBDGCN) to extract both superpixel and pixel-level features for hyperspectral image classification. The TBDGCN model addresses the oversmoothing and overfitting problems by using the DropEdge technique and residual connection in the GCN branch, and captures attention-based spectral-spatial features through a mixed attention mechanism in the CNN branch. The features from both branches are fused for classification.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Environmental Sciences
Yuhao Qing, Wenyi Liu
Summary: A new multi-scale residual convolutional neural network model (MRA-NET) for hyperspectral image classification was proposed, which utilizes an efficient channel attention network to improve deep learning classification accuracy. The evaluation on three public datasets demonstrated higher classification accuracy compared to current networks.
Article
Computer Science, Artificial Intelligence
Xiangtao Zheng, Hao Sun, Xiaoqiang Lu, Wei Xie
Summary: This paper proposes a rotation-invariant attention network (RIAN) for HSI classification, which extracts rotation-invariant spectral-spatial features using center spectral attention and rectified spatial attention modules. Experimental results show that RIAN performs well on HSIs with spatial rotation.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2022)
Article
Computer Science, Information Systems
Wenping Ma, Haoxiang Ma, Hao Zhu, Yating Li, Longwei Li, Licheng Jiao, Biao Hou
Summary: This paper proposes a spatial and spectral kernels generation network (SSKNet) to tackle the classification of hyperspectral images, achieving more efficient feature extraction and fusion through the generation of spatial and spectral convolution kernels. Experimental results demonstrate that this method outperforms existing ones in terms of classification accuracy and generalization performance.
INFORMATION SCIENCES
(2021)
Article
Engineering, Electrical & Electronic
Moqi Liu, Haizhu Pan, Haimiao Ge, Liguo Wang
Summary: This study proposes a novel multiscale stratified-split symmetric network with quadra-view attention (MS3Net) for hyperspectral image (HSI) classification. The MS3Net has a dual-stream symmetric pipeline to extract spectral signatures and spatial features effectively. It consists of three modules: a multiscale feature extraction module, a feature enhancement module, and a feature fusion module. Experimental results demonstrate the superiority of MS3Net over existing methods in terms of visual and quantitative performance on four benchmark HSI datasets.
Article
Environmental Sciences
Dongxu Liu, Yirui Wang, Peixun Liu, Qingqing Li, Hang Yang, Dianbing Chen, Zhichao Liu, Guangliang Han
Summary: This article proposes a multiscale cross interaction attention network (MCIANet) for hyperspectral image classification. It highlights the distinguishability of HSI and dispels redundant information through an interaction attention module (IAM), and detects spectral-spatial features at different scales, convolutional layers, and branches through a multiscale cross feature extraction module (MCFEM). The results show that the presented method outperforms state-of-the-art methods.
Article
Environmental Sciences
Hang Gong, Qiuxia Li, Chunlai Li, Haishan Dai, Zhiping He, Wenjing Wang, Haoyang Li, Feng Han, Abudusalamu Tuniyazi, Tingkui Mu
Summary: Hyperspectral imaging is widely used in classification due to its rich spectral information. This study introduces a lightweight multiscale squeeze-and-excitation pyramid pooling network (MSPN) to address the small sample problem in hyperspectral classification. The MSPN framework can learn and fuse deeper hierarchical spatial-spectral features with fewer training samples, achieving high classification accuracy on various datasets.
Article
Computer Science, Interdisciplinary Applications
Xiaoxia Zhang, Yong Guo, Xia Zhang
Summary: In this study, a 3D CNN network based on stacked blocks was proposed for HSI classification. The proposed network includes an attention mechanism to filter out interfering information. The optimized architecture achieved higher classification rates compared to related works and demonstrated effectiveness and adaptability on a more complex dataset.
EARTH SCIENCE INFORMATICS
(2022)
Article
Environmental Sciences
Haizhu Pan, Xiaoyu Zhao, Haimiao Ge, Moqi Liu, Cuiping Shi
Summary: Hyperspectral image (HSI) classification is a crucial task in remote sensing processing. Attention mechanisms have been favored over convolutional neural networks (CNNs) due to their superior ability to express information during HSI processing. However, achieving high-accuracy classification with limited labeled samples remains a challenge.
Article
Environmental Sciences
Haoyu Jing, Yuanyuan Wang, Zhenhong Du, Feng Zhang
Summary: In this paper, a multi-scale fusion-evolution graph convolutional network based on the feature-spatial attention mechanism is proposed for hyperspectral image classification. Experimental results show that the proposed method outperforms most existing HSI classification methods.
Article
Environmental Sciences
Cuiping Shi, Jingwei Sun, Liguo Wang
Summary: This paper proposes a spatial-spectral attention fusion network for hyperspectral image classification, which uses a four branch multiscale block and 3D-Softpool to extract and fuse spectral and spatial features. Experimental results show that the proposed method outperforms other classification methods when using a small number of training samples.
Article
Environmental Sciences
Cuiping Shi, Xinlei Zhang, Tianyi Wang, Liguo Wang
Summary: This study proposes a lightweight convolutional neural network based on hierarchical-wise convolution fusion for remote-sensing scene image classification. By extracting shallow and deep features, and using global average pooling and Softmax function for classification, it achieves good classification results.
Article
Environmental Sciences
Cuiping Shi, Jingwei Sun, Tianyi Wang, Liguo Wang
Summary: Convolutional neural networks have been widely used in hyperspectral image classification and have achieved good performance. However, the high dimension of hyperspectral images and limited training samples pose challenges. To address this, the paper proposes a new method that combines several attention modules to extract spectral and spatial features and achieves superior classification performance with limited training samples.
Article
Environmental Sciences
Haizhu Pan, Xiaoyu Zhao, Haimiao Ge, Moqi Liu, Cuiping Shi
Summary: Hyperspectral image (HSI) classification is a crucial task in remote sensing processing. Attention mechanisms have been favored over convolutional neural networks (CNNs) due to their superior ability to express information during HSI processing. However, achieving high-accuracy classification with limited labeled samples remains a challenge.