Periocular region-based person identification in the visible, infrared and hyperspectral imagery

M. Uzair, Arif Mahmood, Ajmal Mian, Chris Mcdonald

    Research output: Contribution to journalArticle

    16 Citations (Scopus)
    317 Downloads (Pure)

    Abstract

    Face recognition performance degrades significantly under occlusions that occur intentionally or unintentionally due to head gear or hair style. In many incidents captured by surveillance videos, the offenders cover their faces leaving only the periocular region visible. We present an extensive study on periocular region based person identification in video. While, previous techniques have handpicked a single best frame from videos, we formulate, for the first time, periocular region based person identification in video as an image-set classification problem. For thorough analysis, we perform experiments on periocular regions extracted automatically from RGB videos, NIR videos and hyperspectral image cubes. Each image-set is represented by four heterogeneous feature types and classified with six state-of-the-art image-set classification algorithms. We propose a novel two stage inverse Error Weighted Fusion algorithm for feature and classifier score fusion. The proposed two stage fusion is superior to single stage fusion. Comprehensive experiments were performed on four standard datasets, MBGC NIR and visible spectrum (Phillips et al., 2005), CMU Hyperspectral (Denes et al., 2002) and UBIPr (Padole and Proenca, 2012). We obtained average rank-1 recognition rates of 99.8, 98.5, 97.2, and 99.5% respectively which are significantly higher than the existing state of the art. Our results demonstrate the feasibility of image-set based periocular biometrics for real world applications.
    Original languageEnglish
    Pages (from-to)854-867
    JournalNeurocomputing
    Volume149
    Issue numberPart B
    DOIs
    Publication statusPublished - 3 Feb 2015

    Fingerprint

    Imagery (Psychotherapy)
    Fusion reactions
    Infrared radiation
    Hair
    Head
    Biometrics
    Face recognition
    Gears
    Classifiers
    Experiments
    Facial Recognition
    Datasets

    Cite this

    @article{11e342c178024b1496f0e05f65317862,
    title = "Periocular region-based person identification in the visible, infrared and hyperspectral imagery",
    abstract = "Face recognition performance degrades significantly under occlusions that occur intentionally or unintentionally due to head gear or hair style. In many incidents captured by surveillance videos, the offenders cover their faces leaving only the periocular region visible. We present an extensive study on periocular region based person identification in video. While, previous techniques have handpicked a single best frame from videos, we formulate, for the first time, periocular region based person identification in video as an image-set classification problem. For thorough analysis, we perform experiments on periocular regions extracted automatically from RGB videos, NIR videos and hyperspectral image cubes. Each image-set is represented by four heterogeneous feature types and classified with six state-of-the-art image-set classification algorithms. We propose a novel two stage inverse Error Weighted Fusion algorithm for feature and classifier score fusion. The proposed two stage fusion is superior to single stage fusion. Comprehensive experiments were performed on four standard datasets, MBGC NIR and visible spectrum (Phillips et al., 2005), CMU Hyperspectral (Denes et al., 2002) and UBIPr (Padole and Proenca, 2012). We obtained average rank-1 recognition rates of 99.8, 98.5, 97.2, and 99.5{\%} respectively which are significantly higher than the existing state of the art. Our results demonstrate the feasibility of image-set based periocular biometrics for real world applications.",
    author = "M. Uzair and Arif Mahmood and Ajmal Mian and Chris Mcdonald",
    year = "2015",
    month = "2",
    day = "3",
    doi = "10.1016/j.neucom.2014.07.049",
    language = "English",
    volume = "149",
    pages = "854--867",
    journal = "Neurocomputing",
    issn = "0925-2312",
    publisher = "Pergamon",
    number = "Part B",

    }

    Periocular region-based person identification in the visible, infrared and hyperspectral imagery. / Uzair, M.; Mahmood, Arif; Mian, Ajmal; Mcdonald, Chris.

    In: Neurocomputing, Vol. 149, No. Part B, 03.02.2015, p. 854-867.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Periocular region-based person identification in the visible, infrared and hyperspectral imagery

    AU - Uzair, M.

    AU - Mahmood, Arif

    AU - Mian, Ajmal

    AU - Mcdonald, Chris

    PY - 2015/2/3

    Y1 - 2015/2/3

    N2 - Face recognition performance degrades significantly under occlusions that occur intentionally or unintentionally due to head gear or hair style. In many incidents captured by surveillance videos, the offenders cover their faces leaving only the periocular region visible. We present an extensive study on periocular region based person identification in video. While, previous techniques have handpicked a single best frame from videos, we formulate, for the first time, periocular region based person identification in video as an image-set classification problem. For thorough analysis, we perform experiments on periocular regions extracted automatically from RGB videos, NIR videos and hyperspectral image cubes. Each image-set is represented by four heterogeneous feature types and classified with six state-of-the-art image-set classification algorithms. We propose a novel two stage inverse Error Weighted Fusion algorithm for feature and classifier score fusion. The proposed two stage fusion is superior to single stage fusion. Comprehensive experiments were performed on four standard datasets, MBGC NIR and visible spectrum (Phillips et al., 2005), CMU Hyperspectral (Denes et al., 2002) and UBIPr (Padole and Proenca, 2012). We obtained average rank-1 recognition rates of 99.8, 98.5, 97.2, and 99.5% respectively which are significantly higher than the existing state of the art. Our results demonstrate the feasibility of image-set based periocular biometrics for real world applications.

    AB - Face recognition performance degrades significantly under occlusions that occur intentionally or unintentionally due to head gear or hair style. In many incidents captured by surveillance videos, the offenders cover their faces leaving only the periocular region visible. We present an extensive study on periocular region based person identification in video. While, previous techniques have handpicked a single best frame from videos, we formulate, for the first time, periocular region based person identification in video as an image-set classification problem. For thorough analysis, we perform experiments on periocular regions extracted automatically from RGB videos, NIR videos and hyperspectral image cubes. Each image-set is represented by four heterogeneous feature types and classified with six state-of-the-art image-set classification algorithms. We propose a novel two stage inverse Error Weighted Fusion algorithm for feature and classifier score fusion. The proposed two stage fusion is superior to single stage fusion. Comprehensive experiments were performed on four standard datasets, MBGC NIR and visible spectrum (Phillips et al., 2005), CMU Hyperspectral (Denes et al., 2002) and UBIPr (Padole and Proenca, 2012). We obtained average rank-1 recognition rates of 99.8, 98.5, 97.2, and 99.5% respectively which are significantly higher than the existing state of the art. Our results demonstrate the feasibility of image-set based periocular biometrics for real world applications.

    U2 - 10.1016/j.neucom.2014.07.049

    DO - 10.1016/j.neucom.2014.07.049

    M3 - Article

    VL - 149

    SP - 854

    EP - 867

    JO - Neurocomputing

    JF - Neurocomputing

    SN - 0925-2312

    IS - Part B

    ER -