3D face reconstruction from a single non-frontal face image

Naoki Nozawa, Daiki Kuwahara, Shigeo Morishima

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

    Abstract

    A reconstruction of a human face shape from a single image is an important theme for criminal investigation such as recognition of suspected people from surveillance cameras with only a few frames. It is, however, still difficult to recover a face shape from a non-frontal face image. Method using shading cues on a face depends on the lighting circumstance and cannot be adapted to images in which shadows occurs, for example [Kemelmacher et al. 2011]. On the other hand, [Blanz et al. 2004] reconstructed a shape by 3D Morphable Model (3DMM) only with facial feature points. This method, however, requires the pose-wise correspondences of vertices in the model to feature points of input image because a face contour cannot be seen when the facial direction is not the front. In this paper, we propose a method which can reconstruct a facial shape from a non-frontal face image only with a single general correspondence table. Our method searches for the correspondences of points on a facial contour in the iterative reconstruction process, and makes the reconstruction simple and stable.

    Original languageEnglish
    Title of host publicationACM SIGGRAPH 2015 Posters, SIGGRAPH 2015
    PublisherAssociation for Computing Machinery, Inc
    ISBN (Print)9781450336321
    DOIs
    Publication statusPublished - 2015 Jul 31
    EventInternational Conference on Computer Graphics and Interactive Techniques, SIGGRAPH 2015 - Los Angeles, United States
    Duration: 2015 Aug 92015 Aug 13

    Other

    OtherInternational Conference on Computer Graphics and Interactive Techniques, SIGGRAPH 2015
    CountryUnited States
    CityLos Angeles
    Period15/8/915/8/13

    Fingerprint

    Lighting
    Cameras

    ASJC Scopus subject areas

    • Computer Graphics and Computer-Aided Design
    • Computer Vision and Pattern Recognition
    • Human-Computer Interaction

    Cite this

    Nozawa, N., Kuwahara, D., & Morishima, S. (2015). 3D face reconstruction from a single non-frontal face image. In ACM SIGGRAPH 2015 Posters, SIGGRAPH 2015 [a57] Association for Computing Machinery, Inc. https://doi.org/10.1145/2787626.2792634

    3D face reconstruction from a single non-frontal face image. / Nozawa, Naoki; Kuwahara, Daiki; Morishima, Shigeo.

    ACM SIGGRAPH 2015 Posters, SIGGRAPH 2015. Association for Computing Machinery, Inc, 2015. a57.

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

    Nozawa, N, Kuwahara, D & Morishima, S 2015, 3D face reconstruction from a single non-frontal face image. in ACM SIGGRAPH 2015 Posters, SIGGRAPH 2015., a57, Association for Computing Machinery, Inc, International Conference on Computer Graphics and Interactive Techniques, SIGGRAPH 2015, Los Angeles, United States, 15/8/9. https://doi.org/10.1145/2787626.2792634
    Nozawa N, Kuwahara D, Morishima S. 3D face reconstruction from a single non-frontal face image. In ACM SIGGRAPH 2015 Posters, SIGGRAPH 2015. Association for Computing Machinery, Inc. 2015. a57 https://doi.org/10.1145/2787626.2792634
    Nozawa, Naoki ; Kuwahara, Daiki ; Morishima, Shigeo. / 3D face reconstruction from a single non-frontal face image. ACM SIGGRAPH 2015 Posters, SIGGRAPH 2015. Association for Computing Machinery, Inc, 2015.
    @inproceedings{6a6324ecd17547c0b4c0345ad3ba9154,
    title = "3D face reconstruction from a single non-frontal face image",
    abstract = "A reconstruction of a human face shape from a single image is an important theme for criminal investigation such as recognition of suspected people from surveillance cameras with only a few frames. It is, however, still difficult to recover a face shape from a non-frontal face image. Method using shading cues on a face depends on the lighting circumstance and cannot be adapted to images in which shadows occurs, for example [Kemelmacher et al. 2011]. On the other hand, [Blanz et al. 2004] reconstructed a shape by 3D Morphable Model (3DMM) only with facial feature points. This method, however, requires the pose-wise correspondences of vertices in the model to feature points of input image because a face contour cannot be seen when the facial direction is not the front. In this paper, we propose a method which can reconstruct a facial shape from a non-frontal face image only with a single general correspondence table. Our method searches for the correspondences of points on a facial contour in the iterative reconstruction process, and makes the reconstruction simple and stable.",
    author = "Naoki Nozawa and Daiki Kuwahara and Shigeo Morishima",
    year = "2015",
    month = "7",
    day = "31",
    doi = "10.1145/2787626.2792634",
    language = "English",
    isbn = "9781450336321",
    booktitle = "ACM SIGGRAPH 2015 Posters, SIGGRAPH 2015",
    publisher = "Association for Computing Machinery, Inc",

    }

    TY - GEN

    T1 - 3D face reconstruction from a single non-frontal face image

    AU - Nozawa, Naoki

    AU - Kuwahara, Daiki

    AU - Morishima, Shigeo

    PY - 2015/7/31

    Y1 - 2015/7/31

    N2 - A reconstruction of a human face shape from a single image is an important theme for criminal investigation such as recognition of suspected people from surveillance cameras with only a few frames. It is, however, still difficult to recover a face shape from a non-frontal face image. Method using shading cues on a face depends on the lighting circumstance and cannot be adapted to images in which shadows occurs, for example [Kemelmacher et al. 2011]. On the other hand, [Blanz et al. 2004] reconstructed a shape by 3D Morphable Model (3DMM) only with facial feature points. This method, however, requires the pose-wise correspondences of vertices in the model to feature points of input image because a face contour cannot be seen when the facial direction is not the front. In this paper, we propose a method which can reconstruct a facial shape from a non-frontal face image only with a single general correspondence table. Our method searches for the correspondences of points on a facial contour in the iterative reconstruction process, and makes the reconstruction simple and stable.

    AB - A reconstruction of a human face shape from a single image is an important theme for criminal investigation such as recognition of suspected people from surveillance cameras with only a few frames. It is, however, still difficult to recover a face shape from a non-frontal face image. Method using shading cues on a face depends on the lighting circumstance and cannot be adapted to images in which shadows occurs, for example [Kemelmacher et al. 2011]. On the other hand, [Blanz et al. 2004] reconstructed a shape by 3D Morphable Model (3DMM) only with facial feature points. This method, however, requires the pose-wise correspondences of vertices in the model to feature points of input image because a face contour cannot be seen when the facial direction is not the front. In this paper, we propose a method which can reconstruct a facial shape from a non-frontal face image only with a single general correspondence table. Our method searches for the correspondences of points on a facial contour in the iterative reconstruction process, and makes the reconstruction simple and stable.

    UR - http://www.scopus.com/inward/record.url?scp=84959330987&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84959330987&partnerID=8YFLogxK

    U2 - 10.1145/2787626.2792634

    DO - 10.1145/2787626.2792634

    M3 - Conference contribution

    AN - SCOPUS:84959330987

    SN - 9781450336321

    BT - ACM SIGGRAPH 2015 Posters, SIGGRAPH 2015

    PB - Association for Computing Machinery, Inc

    ER -