Learning Disentangled Priors for Hyperspectral Anomaly Detection: A Coupling Model-Driven and Data-Driven Paradigm

被引:5
|
作者
Li, Chenyu [1 ]
Zhang, Bing [1 ,2 ]
Hong, Danfeng [1 ,3 ]
Jia, Xiuping [4 ]
Plaza, Antonio [5 ]
Chanussot, Jocelyn [1 ,6 ]
机构
[1] Chinese Acad Sci, Aerosp Informat Res Inst, Beijing 100094, Peoples R China
[2] Univ Chinese Acad Sci, Coll Resources & Environm, Beijing 100049, Peoples R China
[3] Univ Chinese Acad Sci, Sch Elect Elect & Commun Engn, Beijing 100049, Peoples R China
[4] Univ New South Wales, Sch Engn & Informat Technol, Canberra, ACT 2612, Australia
[5] Univ Extremadura, Escuela Polit, Dept Technol Comp & Commun, Hyperspectral Comp Lab, Caceres 10003, Spain
[6] Univ Grenoble Alpes, Inria, CNRS, Grenoble INP, F-38000 Grenoble, France
关键词
Task analysis; Mathematical models; Image reconstruction; Data models; Couplings; Hyperspectral imaging; Anomaly detection; disentangled priors; hyperspectral remote sensing; implicit prior; interpretability; low-rank representation (LRR); spatial-spectral attention; LOW-RANK; RX-ALGORITHM; REPRESENTATION;
D O I
10.1109/TNNLS.2024.3401589
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Accurately distinguishing between background and anomalous objects within hyperspectral images poses a significant challenge. The primary obstacle lies in the inadequate modeling of prior knowledge, leading to a performance bottleneck in hyperspectral anomaly detection (HAD). In response to this challenge, we put forth a groundbreaking coupling paradigm that combines model-driven low-rank representation (LRR) methods with data-driven deep learning techniques by learning disentangled priors (LDP). LDP seeks to capture complete priors for effectively modeling the background, thereby extracting anomalies from hyperspectral images more accurately. LDP follows a model-driven deep unfolding architecture, where the prior knowledge is separated into the explicit low-rank prior formulated by expert knowledge and implicit learnable priors by means of deep networks. The internal relationships between explicit and implicit priors within LDP are elegantly modeled through a skip residual connection. Furthermore, we provide a mathematical proof of the convergence of our proposed model. Our experiments, conducted on multiple widely recognized datasets, demonstrate that LDP surpasses most of the current advanced HAD techniques, exceling in both detection performance and generalization capability.
引用
收藏
页数:14
相关论文
共 50 条
  • [1] Hyperspectral Image Denoising: From Model-Driven, Data-Driven, to Model-Data-Driven
    Zhang, Qiang
    Zheng, Yaming
    Yuan, Qiangqiang
    Song, Meiping
    Yu, Haoyang
    Xiao, Yi
    [J]. IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, 2023, : 1 - 21
  • [2] Data-Driven vs Model-Driven Imitative Learning
    Tembine, Hamidou
    [J]. 2017 6TH DATA DRIVEN CONTROL AND LEARNING SYSTEMS (DDCLS), 2017, : 22 - 29
  • [3] An investigation on the coupling of data-driven computing and model-driven computing
    Yang, Jie
    Huang, Wei
    Huang, Qun
    Hu, Heng
    [J]. COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING, 2022, 393
  • [4] A Model-Driven Deep Mixture Network for Robust Hyperspectral Anomaly Detection
    Li, Yunsong
    Jiang, Kai
    Xie, Weiying
    Lei, Jie
    Zhang, Xin
    Du, Qian
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2023, 61
  • [5] Data-driven and Model-driven Deep Learning Detection for RIS-aided Spatial Modulation
    Liu, Jiang
    Di Renzo, Marco
    [J]. 2021 IEEE 4TH 5G WORLD FORUM (5GWF 2021), 2021, : 88 - 92
  • [6] Combining Data-Driven and Model-Driven Methods for Robust Facial Landmark Detection
    Zhang, Hongwen
    Li, Qi
    Sun, Zhenan
    Liu, Yunfan
    [J]. IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, 2018, 13 (10) : 2409 - 2422
  • [7] COMBINED THE DATA-DRIVEN WITH MODEL-DRIVEN STRAGEGY: A NOVEL FRAMEWORK FOR MIXED NOISE REMOVAL IN HYPERSPECTRAL IMAGE
    Zhang, Qiang
    Sun, Fujun
    Yuan, Qiangqiang
    Li, Jie
    Shen, Huanfeng
    Zhang, Liangpei
    [J]. IGARSS 2020 - 2020 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, 2020, : 2667 - 2670
  • [8] On the coupling of data-driven computing and model-driven computing with CUF-based beam models
    Hui, Yanchuan
    Giunta, Gaetano
    Liu, Xiao
    Zheng, Jie
    Yang, Jie
    Huang, Wei
    Carrera, Erasmo
    [J]. THIN-WALLED STRUCTURES, 2024, 195
  • [9] Data-Driven Anomaly Detection for UAV Sensor Data Based on Deep Learning Prediction Model
    Wang, Benkuan
    Wang, Zeyang
    Liu, Liansheng
    Liu, Datong
    Peng, Xiyuan
    [J]. 2019 PROGNOSTICS AND SYSTEM HEALTH MANAGEMENT CONFERENCE (PHM-PARIS), 2019, : 286 - 290
  • [10] A Model-Driven Deep Dehazing Approach by Learning Deep Priors
    Yang, Dong
    Sun, Jian
    [J]. IEEE ACCESS, 2021, 9 : 108542 - 108556