Adversarially Robust Hyperspectral Image Classification via Random Spectral Sampling and Spectral Shape Encoding

被引:10
|
作者
Park, Sungjune [1 ]
Lee, Hong Joo [1 ]
Ro, Yong Man [1 ]
机构
[1] Korea Adv Inst Sci & Technol KAIST, Sch Elect Engn, Daejeon 34141, South Korea
关键词
Training; Robustness; Spectral shape; Shape; Encoding; Feature extraction; Hyperspectral imaging; Adversarial robustness; hyperspectral image classification; random spectral sampling; spectral shape encoding; NEURAL-NETWORKS; ATTACKS; THREAT;
D O I
10.1109/ACCESS.2021.3076225
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Although the hyperspectral image (HSI) classification has adopted deep neural networks (DNNs) and shown remarkable performances, there is a lack of studies of the adversarial vulnerability for the HSI classifications. In this paper, we propose a novel HSI classification framework robust to adversarial attacks. To this end, we focus on the unique spectral characteristic of HSIs (i.e., distinctive spectral patterns of materials). With the spectral characteristic, we present the random spectral sampling and spectral shape feature encoding for the robust HSI classification. For the random spectral sampling, spectral bands are randomly sampled from the entire spectrum for each pixel of the input HSI. Also, the overall spectral shape information, which is robust to adversarial attacks, is fed into the shape feature extractor to acquire the spectral shape feature. Then, the proposed framework can provide the adversarial robustness of HSI classifiers via randomization effects and spectral shape feature encoding. To the best of our knowledge, the proposed framework is the first work dealing with the adversarial robustness in the HSI classification. In experiments, we verify that our framework improves the adversarial robustness considerably under diverse adversarial attack scenarios, and outperforms the existing adversarial defense methods.
引用
收藏
页码:66791 / 66804
页数:14
相关论文
共 50 条
  • [21] Spatial-spectral classification of hyperspectral image via group tensor decomposition
    Zhao, Guangzhe
    Tu, Bing
    Fei, Hongyan
    Li, Nanying
    Yang, Xianchang
    NEUROCOMPUTING, 2018, 316 : 68 - 77
  • [22] Supervised Spectral-Spatial Hyperspectral Image Classification With Weighted Markov Random Fields
    Sun, Le
    Wu, Zebin
    Liu, Jianjun
    Xiao, Liang
    Wei, Zhihui
    IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2015, 53 (03): : 1490 - 1503
  • [23] Spectral-Spatial Hyperspectral Image Classification Using Cascaded Markov Random Fields
    Cao, Xianghai
    Wang, Xiaozhen
    Wang, Da
    Zhao, Jing
    Jiao, Licheng
    IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2019, 12 (12) : 4861 - 4872
  • [24] Multilayer Spectral-Spatial Graphs for Label Noisy Robust Hyperspectral Image Classification
    Jiang, Junjun
    Ma, Jiayi
    Liu, Xianming
    IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2022, 33 (02) : 839 - 852
  • [25] Genie pro: Robust image classification using shape, texture and spectral information
    Perkins, S
    Edlund, K
    Esch-Mosher, D
    Eads, D
    Harvey, N
    Brumby, S
    ALGORITHMS AND TECHNOLOGIES FOR MULTISPECTRAL, HYPERSPECTRAL, AND ULTRASPECTRAL IMAGERY XI, 2005, 5806 : 139 - 148
  • [26] Encoding Spectral-Spatial Features for Hyperspectral Image Classification in the Satellite Internet of Things System
    Lv, Ning
    Han, Zhen
    Chen, Chen
    Feng, Yijia
    Su, Tao
    Goudos, Sotirios
    Wan, Shaohua
    REMOTE SENSING, 2021, 13 (18)
  • [27] Self-supervised spectral clustering with spectral embedding for hyperspectral image classification
    Wu, Chengmao
    Zhang, Jiale
    INTERNATIONAL JOURNAL OF REMOTE SENSING, 2024, 45 (12) : 3913 - 3936
  • [28] Spectral group attention networks for hyperspectral image classification with spectral separability analysis
    Liu, Qi
    Li, Zhengtao
    Shuai, Shuai
    Sun, Qizhen
    INFRARED PHYSICS & TECHNOLOGY, 2020, 108
  • [29] SPECTRAL-SPATIAL HYPERSPECTRAL CLASSIFICATION VIA SHAPE-ADAPTIVE SPARSE REPRESENTATION
    Fu, Wei
    Li, Shutao
    Fang, Leyuan
    Kang, Xudong
    Benediktsson, Jon Atli
    2014 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM (IGARSS), 2014, : 3430 - 3433
  • [30] Local and Global Spectral Features for Hyperspectral Image Classification
    Xu, Zeyu
    Su, Cheng
    Wang, Shirou
    Zhang, Xiaocan
    REMOTE SENSING, 2023, 15 (07)