KECANet: A novel convolutional kernel network for ocean SAR scene classification with limited labeled data

被引:3
|
作者
Ma, Ming [1 ,2 ]
Bai, Chengzu [1 ,2 ]
Zhang, Shuo [1 ,2 ]
Qian, Longxia [3 ]
Yan, Hengqian [4 ]
机构
[1] Beijing Inst Appl Meteorol, Beijing, Peoples R China
[2] State Key Lab Geo Informat Engn, Xian, Peoples R China
[3] Nanjing Univ Posts & Telecommun, Sch Sci, Nanjing, Peoples R China
[4] Natl Univ Def Technol, Coll Meteorol & Oceanog, Changsha, Peoples R China
基金
中国国家自然科学基金;
关键词
synthetic aperture radar; kernel entropy component analysis; convolutional kernel network; oceanic phenomena detection; cholesky decomposition; COMPONENT ANALYSIS; NEURAL-NETWORK; FUSION; CNN;
D O I
10.3389/fmars.2022.935600
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
The ability of synthetic aperture radar (SAR) to capture maritime phenomena is widely acknowledged. However, ocean SAR scene automatic classification remains challenging due to speckle noise interference, the nonlinearities and poor distinguishability of different geophysical phenomena. Kernel entropy component analysis (KECA) was recently proposed as a feature extraction approach. It is capable of handling nonlinear data and revealing different structures of interest. However, KECA suffers from high computational complexity, meaning it cannot penetrate deep for finer feature extraction. To address this issue, this paper proposes an efficient multilayer convolutional kernel network (denoted as KECANet) equipped with KECA for ocean SAR scene classification. The pivoted Cholesky decomposition is employed to accelerate KECA filtering in the network. KECA was trained on hand-labeled but limited samples describing ten oceanic or atmospheric phenomena. Several conventional and state-of-the-art deep learning methods were also included for comparison. According to the classification experiments, KECANet can greatly improve the classification precision of geophysical phenomena, considering that the precision, recall and F-score values increased by 13.3%, 2.3% and 12.2% in average. Overall, the results suggest that KECA is a promising approach for various applications in remote sensing image recognition.
引用
收藏
页数:16
相关论文
共 50 条
  • [41] A Joint Convolutional Neural Network for Simultaneous Despeckling and Classification of SAR Targets
    Lei, Peng
    Zheng, Tong
    Wang, Jun
    Bai, Xiao
    IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2021, 18 (09) : 1610 - 1614
  • [42] Polarimetric SAR Targets Detection and Classification with Deep Convolutional Neural Network
    Chen, Si-Wei
    Tao, Chen-Song
    Wang, Xue-Song
    Xiao, Shun-Ping
    2018 PROGRESS IN ELECTROMAGNETICS RESEARCH SYMPOSIUM (PIERS-TOYAMA), 2018, : 2227 - 2234
  • [43] SAR image classification with convolutional neural network using modified functions
    Soltanali, Aliasghar
    Ghods, Vahid
    Mousavizadeh, Seyed Farhood
    Amirahmadi, Meysam
    SOFT COMPUTING, 2023, 28 (7-8) : 6039 - 6057
  • [44] Hyperspectral and SAR Image Classification via Graph Convolutional Fusion Network
    Deng, Bin
    Duan, Puhong
    Lu, Xukun
    Wang, Zihao
    Kang, Xudong
    IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2024, 62
  • [45] FusioNet: A Two-Stream Convolutional Neural Network for Urban Scene Classification using PolSAR and Hyperspectral Data
    Hu, Jingliang
    Mou, Lichao
    Schmitt, Andreas
    Zhu, Xiao Xiang
    2017 JOINT URBAN REMOTE SENSING EVENT (JURSE), 2017,
  • [46] Scene classification of multisource remote sensing data with two-stream densely connected convolutional neural network
    Wan, Zhen
    Yang, Ronghua
    You, Yangsheng
    Cao, Zhilin
    Fang, Xinan
    IMAGE AND SIGNAL PROCESSING FOR REMOTE SENSING XXIV, 2018, 10789
  • [47] A Novel Convolutional Neural Network Architecture for SAR Target Recognition
    Xie, Yinjie
    Dai, Wenxin
    Hu, Zhenxin
    Liu, Yijing
    Li, Chuan
    Pu, Xuemei
    JOURNAL OF SENSORS, 2019, 2019
  • [48] SAR TARGET CLASSIFICATION WITH LIMITED DATA VIA DATA DRIVEN ACTIVE LEARNING
    Zhou, Yue
    Jiang, Xue
    Li, Zhou
    Liu, Xingzhao
    IGARSS 2020 - 2020 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, 2020, : 2475 - 2478
  • [49] Convolutional Neural Network With Data Augmentation for SAR Target Recognition
    Ding, Jun
    Chen, Bo
    Liu, Hongwei
    Huang, Mengyuan
    IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2016, 13 (03) : 364 - 368
  • [50] CFCANet: A Complete Frequency Channel Attention Network for SAR Image Scene Classification
    Su, Bo
    Liu, Jun
    Su, Xin
    Luo, Bin
    Wang, Qing
    IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2021, 14 : 11750 - 11763