A Hierarchical Coarse-Fine Adaptive Fusion Network for the Joint Classification of Hyperspectral and LiDAR Data

被引:0
作者
Pan, Haizhu [1 ,2 ]
Li, Xuan [1 ]
Ge, Haimiao [1 ,2 ]
Wang, Liguo [3 ]
Shi, Cuiping [4 ]
机构
[1] Qiqihar Univ, Coll Comp & Control Engn, Qiqihar 161000, Peoples R China
[2] Qiqihar Univ, Heilongjiang Key Lab Big Data Network Secur Detect, Qiqihar 161000, Peoples R China
[3] Dalian Nationalities Univ, Coll Informat & Commun Engn, Dalian 116000, Peoples R China
[4] Huzhou Univ, Coll Informat Engn, Huzhou 313000, Peoples R China
基金
中国国家自然科学基金;
关键词
coarse-fine grained features; dynamic convolution; transformer; mutual learning; multi-source; LAND-COVER CLASSIFICATION; IMAGE CLASSIFICATION;
D O I
10.3390/rs16214029
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Hyperspectral image (HSI) and light detection and ranging (LiDAR) data have gained significant attention due to their excellent complementarity, which can be combined to improve the accuracy of land cover classification. However, there are still many challenges, such as unrelated physical characteristics, different data structures, and a lack of labeled samples. Many methods fail to leverage the full potential of multi-source data, particularly hierarchical complementary information. To address these problems, a hierarchical coarse-fine adaptive (HCFA) fusion network with dynamic convolution and a transformer is proposed for multi-source remote sensing land cover classification. Although fusing hierarchical information can improve the classification accuracy of the model, improper hierarchical feature selection and optimization may negatively affect the classification results. Therefore, a coarse-fine mutual learning strategy is proposed to dynamically fuse hierarchical information. Additionally, the disparity between multi-source data continues to prevent the realization of effective fusion. To tackle this challenge, cross-tokenization and cross-token attention are implemented to enhance information interaction. Furthermore, to improve the model representation with limited computational cost, we combine the advantages of dynamic convolution with a transformer. Validation on three standard datasets demonstrates that HCFA achieves high accuracy with just 1% of the training set while maintaining low computational costs.
引用
收藏
页数:28
相关论文
共 50 条
[21]   Deep Hierarchical Vision Transformer for Hyperspectral and LiDAR Data Classification [J].
Xue, Zhixiang ;
Tan, Xiong ;
Yu, Xuchu ;
Liu, Bing ;
Yu, Anzhu ;
Zhang, Pengqiang .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2022, 31 :3095-3110
[22]   DISCRIMINATIVE FEATURE EXTRACTION AND FUSION FOR CLASSIFICATION OF HYPERSPECTRAL AND LIDAR DATA [J].
Song, Weiwei ;
Gao, Zhi ;
Zhang, Yongjun .
2022 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM (IGARSS 2022), 2022, :2271-2274
[23]   Attention-Guided Fusion and Classification for Hyperspectral and LiDAR Data [J].
Huang, Jing ;
Zhang, Yinghao ;
Yang, Fang ;
Chai, Li ;
Tansey, Kevin .
REMOTE SENSING, 2024, 16 (01)
[24]   Fusion of Hyperspectral CASI and Airborne LiDAR Data for Ground Object Classification through Residual Network [J].
Chang, Zhanyuan ;
Yu, Huiling ;
Zhang, Yizhuo ;
Wang, Keqi .
SENSORS, 2020, 20 (14) :1-16
[25]   A NEW MULTI-LEVEL ATTENTION FEATURE FUSION METHOD FOR HYPERSPECTRAL AND LIDAR DATA JOINT CLASSIFICATION [J].
Song, Weiwei ;
Gao, Zhi ;
Fang, Leyuan ;
Zhang, Yongjun .
IGARSS 2023 - 2023 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, 2023, :5978-5981
[26]   Spatial-Spectral Heterogeneity-Aware Network for Hyperspectral and LiDAR Joint Classification [J].
Zhang, Shenfu ;
Liu, Qiang ;
Zhang, Zhenhua ;
Zhao, Rui ;
Chen, Liang ;
Shao, Feng ;
Meng, Xiangchao .
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2025,
[27]   MCFNet: Multiscale Cross-Domain Fusion Network for HSI and LiDAR Data Joint Classification [J].
Song, Qiya ;
Mo, Feng ;
Ding, Kexin ;
Xiao, Lin ;
Dian, Renwei ;
Kang, Xudong ;
Li, Shutao .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2025, 63
[28]   Feature-Decision Level Collaborative Fusion Network for Hyperspectral and LiDAR Classification [J].
Zhang, Shenfu ;
Meng, Xiangchao ;
Liu, Qiang ;
Yang, Gang ;
Sun, Weiwei .
REMOTE SENSING, 2023, 15 (17)
[29]   Center-bridged Interaction Fusion for hyperspectral and LiDAR classification [J].
Huo, Lu ;
Xia, Jiahao ;
Zhang, Leijie ;
Zhang, Haimin ;
Xu, Min .
NEUROCOMPUTING, 2024, 590
[30]   A Joint Convolutional Cross ViT Network for Hyperspectral and Light Detection and Ranging Fusion Classification [J].
Xu, Haitao ;
Zheng, Tie ;
Liu, Yuzhe ;
Zhang, Zhiyuan ;
Xue, Changbin ;
Li, Jiaojiao .
REMOTE SENSING, 2024, 16 (03)