SuperPCA: A Superpixelwise PCA Approach for Unsupervised Feature Extraction of Hyperspectral Imagery

被引:294
作者
Jiang, Junjun [1 ]
Ma, Jiayi [2 ]
Chen, Chen [3 ]
Wang, Zhongyuan [4 ]
Cai, Zhihua [5 ]
Wang, Lizhe [5 ]
机构
[1] Harbin Inst Technol, Sch Comp Sci & Technol, Harbin 150001, Heilongjiang, Peoples R China
[2] Wuhan Univ, Elect Informat Sch, Wuhan 430072, Hubei, Peoples R China
[3] Univ North Carolina Charlotte, Dept Elect & Comp Engn, Charlotte, NC 28223 USA
[4] Wuhan Univ, Sch Comp, Wuhan 430072, Hubei, Peoples R China
[5] China Univ Geosci, Sch Comp Sci, Wuhan 430074, Hubei, Peoples R China
来源
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING | 2018年 / 56卷 / 08期
基金
中国国家自然科学基金;
关键词
Feature extraction; hyperspectral image (HSI) classification; principal component analysis (PCA); superpixel segmentation; unsupervised dimensionality reduction; DIMENSIONALITY REDUCTION; SPARSE REPRESENTATION; DISCRIMINANT-ANALYSIS; COMPONENT ANALYSIS; CLASSIFICATION; FUSION;
D O I
10.1109/TGRS.2018.2828029
中图分类号
P3 [地球物理学]; P59 [地球化学];
学科分类号
0708 ; 070902 ;
摘要
As an unsupervised dimensionality reduction method, the principal component analysis (PCA) has been widely considered as an efficient and effective preprocessing step for hyperspectral image (HSI) processing and analysis tasks. It takes each band as a whole and globally extracts the most representative hands. However, different homogeneous regions correspond to different objects, whose spectral features are diverse. Therefore, it is inappropriate to carry out dimensionality reduction through a unified projection for an entire HSI. In this paper, a simple but very effective superpixelwise PCA (SuperPCA) approach is proposed to learn the intrinsic low-dimensional features of HSIs. In contrast to classical PCA models, the SuperPCA has four main properties: 1) unlike the traditional PCA method based on a whole image, the SuperPCA takes into account the diversity in different homogeneous regions, that is, different regions should have different projections; 2) most of the conventional feature extraction models cannot directly use the spatial information of HSIs, while the SuperPCA is able to incorporate the spatial context information into the unsupervised dimensionality reduction by superpixel segmentation; 3) since the regions obtained by superpixel segmentation have homogeneity, the SuperPCA can extract potential low-dimensional features even under noise; and 4) although the SuperPCA is an unsupervised method, it can achieve a competitive performance when compared with supervised approaches. The resulting features are discriminative, compact, and noise-resistant, leading to an improved HSI classification performance. Experiments on three public data sets demonstrate that the SuperPCA model significantly outperforms the conventional PCA-based dimensionality reduction baselines for HSI classification, and some state-of-the-art feature extraction approaches. The MATLAB source code is available at https://github.com/junjun-jiang/SuperPCA.
引用
收藏
页码:4581 / 4593
页数:13
相关论文
共 65 条
  • [21] Patch-Based Principal Component Analysis for Face Recognition
    Jiang, Tai-Xiang
    Huang, Ting-Zhu
    Zhao, Xi-Le
    Ma, Tian-Hui
    [J]. COMPUTATIONAL INTELLIGENCE AND NEUROSCIENCE, 2017, 2017
  • [22] Dimensionality Reduction via Regression in Hyperspectral Imagery
    Laparra, Valero
    Malo, Jesus
    Camps-Valls, Gustau
    [J]. IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING, 2015, 9 (06) : 1026 - 1036
  • [23] TurboPixels: Fast Superpixels Using Geometric Flows
    Levinshtein, Alex
    Stere, Adrian
    Kutulakos, Kiriakos N.
    Fleet, David J.
    Dickinson, Sven J.
    Siddiqi, Kaleem
    [J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2009, 31 (12) : 2290 - 2297
  • [24] Efficient Superpixel-Level Multitask Joint Sparse Representation for Hyperspectral Image Classification
    Li, Jiayi
    Zhang, Hongyan
    Zhang, Liangpei
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2015, 53 (10): : 5338 - 5351
  • [25] Sparse and Low-Rank Graph for Discriminant Analysis of Hyperspectral Imagery
    Li, Wei
    Liu, Jiabin
    Du, Qian
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2016, 54 (07): : 4094 - 4105
  • [26] Classification and Reconstruction From Random Projections for Hyperspectral Imagery
    Li, Wei
    Prasad, Saurabh
    Fowler, James E.
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2013, 51 (02): : 833 - 843
  • [27] Locality-Preserving Dimensionality Reduction and Classification for Hyperspectral Image Analysis
    Li, Wei
    Prasad, Saurabh
    Fowler, James E.
    Bruce, Lori Mann
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2012, 50 (04): : 1185 - 1198
  • [28] Manifold-Learning-Based Feature Extraction for Classification of Hyperspectral Data
    Lunga, Dalton
    Prasad, Saurabh
    Crawford, Melba M.
    Ersoy, Okan
    [J]. IEEE SIGNAL PROCESSING MAGAZINE, 2014, 31 (01) : 55 - 66
  • [29] Collaborative Graph-Based Discriminant Analysis for Hyperspectral Imagery
    Ly, Nam Hoai
    Du, Qian
    Fowler, James E.
    [J]. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2014, 7 (06) : 2688 - 2696
  • [30] Sparse Graph-Based Discriminant Analysis for Hyperspectral Imagery
    Ly, Nam Hoai
    Du, Qian
    Fowler, James E.
    [J]. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2014, 52 (07): : 3872 - 3884