Study of recognizing human motion observed from an arbitrary viewpoint based on decomposition of a tensor containing multiple view motions

Takayuki Hori, Jun Ohya, Jun Kurumisawa

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    We propose a Tensor Decomposition based algorithm that recognizes the observed action performed by an unknown person and unknown viewpoint not included in the database. Our previous research aimed motion recognition from one single viewpoint. In this paper, we extend our approach for human motion recognition from an arbitrary viewpoint. To achieve this issue, we set tensor database which are multi-dimensional vectors with dimensions corresponding to human models, viewpoint angles, and action classes. The value of a tensor for a given combination of human silhouette model, viewpoint angle, and action class is the series of mesh feature vectors calculated each frame sequence. To recognize human motion, the actions of one of the persons in the tensor are replaced by the synthesized actions. Then, the core tensor for the replaced tensor is computed. This process is repeated for each combination of action, person, and viewpoint. For each iteration, the difference between the replaced and original core tensors is computed. The assumption that gives the minimal difference is the action recognition result. The recognition results show the validity of our proposed method, the method is experimentally compared with Nearest Neighbor rule. Our proposed method is very stable as each action was recognized with over 75% accuracy.

    Original languageEnglish
    Title of host publicationProceedings of SPIE - The International Society for Optical Engineering
    Volume7873
    DOIs
    Publication statusPublished - 2011
    EventComputational Imaging IX - San Francisco, CA
    Duration: 2011 Jan 242011 Jan 25

    Other

    OtherComputational Imaging IX
    CitySan Francisco, CA
    Period11/1/2411/1/25

    Fingerprint

    Tensors
    Tensor
    tensors
    Decomposition
    decomposition
    Decompose
    Motion
    Arbitrary
    Person
    Tensor Decomposition
    Angle
    Action Recognition
    Unknown
    Silhouette
    Human
    Feature Vector
    iteration
    mesh
    Nearest Neighbor
    Mesh

    Keywords

    • Computer vision
    • Core tensor
    • Human motion analysis
    • Human motion recognition
    • Motion signature
    • Multiple viewpoint
    • N-mode SVD
    • Tensor decomposition

    ASJC Scopus subject areas

    • Applied Mathematics
    • Computer Science Applications
    • Electrical and Electronic Engineering
    • Electronic, Optical and Magnetic Materials
    • Condensed Matter Physics

    Cite this

    Hori, T., Ohya, J., & Kurumisawa, J. (2011). Study of recognizing human motion observed from an arbitrary viewpoint based on decomposition of a tensor containing multiple view motions. In Proceedings of SPIE - The International Society for Optical Engineering (Vol. 7873). [787310] https://doi.org/10.1117/12.872300

    Study of recognizing human motion observed from an arbitrary viewpoint based on decomposition of a tensor containing multiple view motions. / Hori, Takayuki; Ohya, Jun; Kurumisawa, Jun.

    Proceedings of SPIE - The International Society for Optical Engineering. Vol. 7873 2011. 787310.

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Hori, T, Ohya, J & Kurumisawa, J 2011, Study of recognizing human motion observed from an arbitrary viewpoint based on decomposition of a tensor containing multiple view motions. in Proceedings of SPIE - The International Society for Optical Engineering. vol. 7873, 787310, Computational Imaging IX, San Francisco, CA, 11/1/24. https://doi.org/10.1117/12.872300
    Hori T, Ohya J, Kurumisawa J. Study of recognizing human motion observed from an arbitrary viewpoint based on decomposition of a tensor containing multiple view motions. In Proceedings of SPIE - The International Society for Optical Engineering. Vol. 7873. 2011. 787310 https://doi.org/10.1117/12.872300
    Hori, Takayuki ; Ohya, Jun ; Kurumisawa, Jun. / Study of recognizing human motion observed from an arbitrary viewpoint based on decomposition of a tensor containing multiple view motions. Proceedings of SPIE - The International Society for Optical Engineering. Vol. 7873 2011.
    @inproceedings{9306cad759544714a650df43a6e5518a,
    title = "Study of recognizing human motion observed from an arbitrary viewpoint based on decomposition of a tensor containing multiple view motions",
    abstract = "We propose a Tensor Decomposition based algorithm that recognizes the observed action performed by an unknown person and unknown viewpoint not included in the database. Our previous research aimed motion recognition from one single viewpoint. In this paper, we extend our approach for human motion recognition from an arbitrary viewpoint. To achieve this issue, we set tensor database which are multi-dimensional vectors with dimensions corresponding to human models, viewpoint angles, and action classes. The value of a tensor for a given combination of human silhouette model, viewpoint angle, and action class is the series of mesh feature vectors calculated each frame sequence. To recognize human motion, the actions of one of the persons in the tensor are replaced by the synthesized actions. Then, the core tensor for the replaced tensor is computed. This process is repeated for each combination of action, person, and viewpoint. For each iteration, the difference between the replaced and original core tensors is computed. The assumption that gives the minimal difference is the action recognition result. The recognition results show the validity of our proposed method, the method is experimentally compared with Nearest Neighbor rule. Our proposed method is very stable as each action was recognized with over 75{\%} accuracy.",
    keywords = "Computer vision, Core tensor, Human motion analysis, Human motion recognition, Motion signature, Multiple viewpoint, N-mode SVD, Tensor decomposition",
    author = "Takayuki Hori and Jun Ohya and Jun Kurumisawa",
    year = "2011",
    doi = "10.1117/12.872300",
    language = "English",
    isbn = "9780819484109",
    volume = "7873",
    booktitle = "Proceedings of SPIE - The International Society for Optical Engineering",

    }

    TY - GEN

    T1 - Study of recognizing human motion observed from an arbitrary viewpoint based on decomposition of a tensor containing multiple view motions

    AU - Hori, Takayuki

    AU - Ohya, Jun

    AU - Kurumisawa, Jun

    PY - 2011

    Y1 - 2011

    N2 - We propose a Tensor Decomposition based algorithm that recognizes the observed action performed by an unknown person and unknown viewpoint not included in the database. Our previous research aimed motion recognition from one single viewpoint. In this paper, we extend our approach for human motion recognition from an arbitrary viewpoint. To achieve this issue, we set tensor database which are multi-dimensional vectors with dimensions corresponding to human models, viewpoint angles, and action classes. The value of a tensor for a given combination of human silhouette model, viewpoint angle, and action class is the series of mesh feature vectors calculated each frame sequence. To recognize human motion, the actions of one of the persons in the tensor are replaced by the synthesized actions. Then, the core tensor for the replaced tensor is computed. This process is repeated for each combination of action, person, and viewpoint. For each iteration, the difference between the replaced and original core tensors is computed. The assumption that gives the minimal difference is the action recognition result. The recognition results show the validity of our proposed method, the method is experimentally compared with Nearest Neighbor rule. Our proposed method is very stable as each action was recognized with over 75% accuracy.

    AB - We propose a Tensor Decomposition based algorithm that recognizes the observed action performed by an unknown person and unknown viewpoint not included in the database. Our previous research aimed motion recognition from one single viewpoint. In this paper, we extend our approach for human motion recognition from an arbitrary viewpoint. To achieve this issue, we set tensor database which are multi-dimensional vectors with dimensions corresponding to human models, viewpoint angles, and action classes. The value of a tensor for a given combination of human silhouette model, viewpoint angle, and action class is the series of mesh feature vectors calculated each frame sequence. To recognize human motion, the actions of one of the persons in the tensor are replaced by the synthesized actions. Then, the core tensor for the replaced tensor is computed. This process is repeated for each combination of action, person, and viewpoint. For each iteration, the difference between the replaced and original core tensors is computed. The assumption that gives the minimal difference is the action recognition result. The recognition results show the validity of our proposed method, the method is experimentally compared with Nearest Neighbor rule. Our proposed method is very stable as each action was recognized with over 75% accuracy.

    KW - Computer vision

    KW - Core tensor

    KW - Human motion analysis

    KW - Human motion recognition

    KW - Motion signature

    KW - Multiple viewpoint

    KW - N-mode SVD

    KW - Tensor decomposition

    UR - http://www.scopus.com/inward/record.url?scp=79953021850&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=79953021850&partnerID=8YFLogxK

    U2 - 10.1117/12.872300

    DO - 10.1117/12.872300

    M3 - Conference contribution

    SN - 9780819484109

    VL - 7873

    BT - Proceedings of SPIE - The International Society for Optical Engineering

    ER -