Nonlinear PCA/ICA for the structure from motion problem

Jun Fujiki, Shotaro Akaho, Noboru Murata

    Research output: Contribution to journalArticle

    2 Citations (Scopus)

    Abstract

    Recovering both camera motion and object shape from multiple images,' called structure from motion problem, is an important and essential problem in computer vision. Generally, the result of the structure from motion problem has an ambiguity represented by a threedimensional rotation matrix. We present two kinds of specific criteria such as independence of parameters to fix the ambiguity by choosing an appropriate rotation matrix in the sense of computer vision. Once some criterion is defined, the fixing of the ambiguity is reduced to a nonlinear extension of the PCA/ICA. We examine the efficiency through synthetic experiments.

    Original languageEnglish
    Pages (from-to)750-757
    Number of pages8
    JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume3195
    Publication statusPublished - 2004

    Fingerprint

    Structure from Motion
    Passive Cutaneous Anaphylaxis
    Independent component analysis
    Computer vision
    Rotation matrix
    Computer Vision
    Cameras
    Camera
    Three-dimensional
    Motion
    Experiments
    Experiment
    Ambiguity

    ASJC Scopus subject areas

    • Computer Science(all)
    • Biochemistry, Genetics and Molecular Biology(all)
    • Theoretical Computer Science

    Cite this

    @article{5f45b4b72a6d4add983c64052aca8a56,
    title = "Nonlinear PCA/ICA for the structure from motion problem",
    abstract = "Recovering both camera motion and object shape from multiple images,' called structure from motion problem, is an important and essential problem in computer vision. Generally, the result of the structure from motion problem has an ambiguity represented by a threedimensional rotation matrix. We present two kinds of specific criteria such as independence of parameters to fix the ambiguity by choosing an appropriate rotation matrix in the sense of computer vision. Once some criterion is defined, the fixing of the ambiguity is reduced to a nonlinear extension of the PCA/ICA. We examine the efficiency through synthetic experiments.",
    author = "Jun Fujiki and Shotaro Akaho and Noboru Murata",
    year = "2004",
    language = "English",
    volume = "3195",
    pages = "750--757",
    journal = "Lecture Notes in Computer Science",
    issn = "0302-9743",
    publisher = "Springer Verlag",

    }

    TY - JOUR

    T1 - Nonlinear PCA/ICA for the structure from motion problem

    AU - Fujiki, Jun

    AU - Akaho, Shotaro

    AU - Murata, Noboru

    PY - 2004

    Y1 - 2004

    N2 - Recovering both camera motion and object shape from multiple images,' called structure from motion problem, is an important and essential problem in computer vision. Generally, the result of the structure from motion problem has an ambiguity represented by a threedimensional rotation matrix. We present two kinds of specific criteria such as independence of parameters to fix the ambiguity by choosing an appropriate rotation matrix in the sense of computer vision. Once some criterion is defined, the fixing of the ambiguity is reduced to a nonlinear extension of the PCA/ICA. We examine the efficiency through synthetic experiments.

    AB - Recovering both camera motion and object shape from multiple images,' called structure from motion problem, is an important and essential problem in computer vision. Generally, the result of the structure from motion problem has an ambiguity represented by a threedimensional rotation matrix. We present two kinds of specific criteria such as independence of parameters to fix the ambiguity by choosing an appropriate rotation matrix in the sense of computer vision. Once some criterion is defined, the fixing of the ambiguity is reduced to a nonlinear extension of the PCA/ICA. We examine the efficiency through synthetic experiments.

    UR - http://www.scopus.com/inward/record.url?scp=35048843747&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=35048843747&partnerID=8YFLogxK

    M3 - Article

    AN - SCOPUS:35048843747

    VL - 3195

    SP - 750

    EP - 757

    JO - Lecture Notes in Computer Science

    JF - Lecture Notes in Computer Science

    SN - 0302-9743

    ER -