RCMF: Robust Constrained Matrix Factorization for Hyperspectral Unmixing

    Research output: Contribution to journalArticle

    8 Citations (Scopus)
    290 Downloads (Pure)

    Abstract

    We propose a constrained matrix factorization approach for linear unmixing of hyperspectral data. Our approach factorizes a hyperspectral cube into its constituent endmembers and their fractional abundances such that the endmembers are sparse nonnegative linear combinations of the observed spectra themselves. The association between the extracted endmembers and the observed spectra is explicitly noted for physical interpretability. To ensure reliable unmixing, we make the matrix factorization procedure robust to outliers in the observed spectra. Our approach simultaneously computes the endmembers and their abundances in an efficient and unsupervised manner. The extracted endmembers are nonnegative quantities, whereas their abundances additionally follow the sum-to-one constraint. We thoroughly evaluate our approach using synthetic data with white and correlated noise as well as real hyperspectral data. Experimental results establish the effectiveness of our approach.

    Original languageEnglish
    Pages (from-to)3354-3366
    Number of pages13
    JournalIEEE Transactions on Geoscience and Remote Sensing
    Volume55
    Issue number6
    DOIs
    Publication statusPublished - Jun 2017

    Cite this

    @article{e16476fda804411c89033880a5008f5e,
    title = "RCMF: Robust Constrained Matrix Factorization for Hyperspectral Unmixing",
    abstract = "We propose a constrained matrix factorization approach for linear unmixing of hyperspectral data. Our approach factorizes a hyperspectral cube into its constituent endmembers and their fractional abundances such that the endmembers are sparse nonnegative linear combinations of the observed spectra themselves. The association between the extracted endmembers and the observed spectra is explicitly noted for physical interpretability. To ensure reliable unmixing, we make the matrix factorization procedure robust to outliers in the observed spectra. Our approach simultaneously computes the endmembers and their abundances in an efficient and unsupervised manner. The extracted endmembers are nonnegative quantities, whereas their abundances additionally follow the sum-to-one constraint. We thoroughly evaluate our approach using synthetic data with white and correlated noise as well as real hyperspectral data. Experimental results establish the effectiveness of our approach.",
    keywords = "Blind source separation, hyperspectral unmixing, robust matrix factorization, sparse representation, unsupervised unmixing, COLLABORATIVE SPARSE REGRESSION, ORTHOGONAL MATCHING PURSUIT, ENDMEMBER EXTRACTION, IMAGING SPECTROSCOPY, ARCHETYPAL ANALYSIS, SIGNAL RECOVERY, ALGORITHM, NMF, REPRESENTATION, DICTIONARIES",
    author = "Naveed Akhtar and Ajmal Mian",
    year = "2017",
    month = "6",
    doi = "10.1109/TGRS.2017.2669991",
    language = "English",
    volume = "55",
    pages = "3354--3366",
    journal = "IEEE Transactions on Geoscience and Remote Sensing",
    issn = "0196-2892",
    publisher = "IEEE, Institute of Electrical and Electronics Engineers",
    number = "6",

    }

    RCMF : Robust Constrained Matrix Factorization for Hyperspectral Unmixing. / Akhtar, Naveed; Mian, Ajmal.

    In: IEEE Transactions on Geoscience and Remote Sensing, Vol. 55, No. 6, 06.2017, p. 3354-3366.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - RCMF

    T2 - Robust Constrained Matrix Factorization for Hyperspectral Unmixing

    AU - Akhtar, Naveed

    AU - Mian, Ajmal

    PY - 2017/6

    Y1 - 2017/6

    N2 - We propose a constrained matrix factorization approach for linear unmixing of hyperspectral data. Our approach factorizes a hyperspectral cube into its constituent endmembers and their fractional abundances such that the endmembers are sparse nonnegative linear combinations of the observed spectra themselves. The association between the extracted endmembers and the observed spectra is explicitly noted for physical interpretability. To ensure reliable unmixing, we make the matrix factorization procedure robust to outliers in the observed spectra. Our approach simultaneously computes the endmembers and their abundances in an efficient and unsupervised manner. The extracted endmembers are nonnegative quantities, whereas their abundances additionally follow the sum-to-one constraint. We thoroughly evaluate our approach using synthetic data with white and correlated noise as well as real hyperspectral data. Experimental results establish the effectiveness of our approach.

    AB - We propose a constrained matrix factorization approach for linear unmixing of hyperspectral data. Our approach factorizes a hyperspectral cube into its constituent endmembers and their fractional abundances such that the endmembers are sparse nonnegative linear combinations of the observed spectra themselves. The association between the extracted endmembers and the observed spectra is explicitly noted for physical interpretability. To ensure reliable unmixing, we make the matrix factorization procedure robust to outliers in the observed spectra. Our approach simultaneously computes the endmembers and their abundances in an efficient and unsupervised manner. The extracted endmembers are nonnegative quantities, whereas their abundances additionally follow the sum-to-one constraint. We thoroughly evaluate our approach using synthetic data with white and correlated noise as well as real hyperspectral data. Experimental results establish the effectiveness of our approach.

    KW - Blind source separation

    KW - hyperspectral unmixing

    KW - robust matrix factorization

    KW - sparse representation

    KW - unsupervised unmixing

    KW - COLLABORATIVE SPARSE REGRESSION

    KW - ORTHOGONAL MATCHING PURSUIT

    KW - ENDMEMBER EXTRACTION

    KW - IMAGING SPECTROSCOPY

    KW - ARCHETYPAL ANALYSIS

    KW - SIGNAL RECOVERY

    KW - ALGORITHM

    KW - NMF

    KW - REPRESENTATION

    KW - DICTIONARIES

    U2 - 10.1109/TGRS.2017.2669991

    DO - 10.1109/TGRS.2017.2669991

    M3 - Article

    VL - 55

    SP - 3354

    EP - 3366

    JO - IEEE Transactions on Geoscience and Remote Sensing

    JF - IEEE Transactions on Geoscience and Remote Sensing

    SN - 0196-2892

    IS - 6

    ER -