An Integrated Framework for 3-D Modeling, Object Detection, and Pose Estimation From Point-Clouds

    Research output: Contribution to journalArticle

    29 Citations (Scopus)

    Abstract

    3-D modeling, object detection, and pose estimation are three of the most challenging tasks in the area of 3-D computer vision. This paper presents a novel algorithm to perform these tasks simultaneously from unordered point-clouds. Given a set of input point-clouds in the presence of clutter and occlusion, an initial model is first constructed by performing pair-wise registration between any two point-clouds. The resulting model is then updated from the remaining point-clouds using a novel model growing technique. Once the final model is reconstructed, the instances of the object are detected and the poses of its instances in the scenes are estimated. This algorithm is automatic, model free, and does not rely on any prior information about the objects in the scene. The algorithm was comprehensively tested on the University of Western Australia data set. Experimental results show that our algorithm achieved accurate modeling, detection, and pose estimation performance.
    Original languageEnglish
    Pages (from-to)683-693
    Number of pages11
    JournalIEEE Transactions on Instrumentation and Measurement
    Volume64
    Issue number3
    DOIs
    Publication statusPublished - Mar 2015

    Fingerprint

    occlusion
    computer vision
    clutter
    Computer vision
    Object detection

    Cite this

    @article{2ff51d4d28a4450ea8d3c00082764880,
    title = "An Integrated Framework for 3-D Modeling, Object Detection, and Pose Estimation From Point-Clouds",
    abstract = "3-D modeling, object detection, and pose estimation are three of the most challenging tasks in the area of 3-D computer vision. This paper presents a novel algorithm to perform these tasks simultaneously from unordered point-clouds. Given a set of input point-clouds in the presence of clutter and occlusion, an initial model is first constructed by performing pair-wise registration between any two point-clouds. The resulting model is then updated from the remaining point-clouds using a novel model growing technique. Once the final model is reconstructed, the instances of the object are detected and the poses of its instances in the scenes are estimated. This algorithm is automatic, model free, and does not rely on any prior information about the objects in the scene. The algorithm was comprehensively tested on the University of Western Australia data set. Experimental results show that our algorithm achieved accurate modeling, detection, and pose estimation performance.",
    author = "Yulan Guo and Mohammed Bennamoun and Ferdous Sohel and M. Lu and J. Wan",
    year = "2015",
    month = "3",
    doi = "10.1109/TIM.2014.2358131",
    language = "English",
    volume = "64",
    pages = "683--693",
    journal = "Institute of Electrical and Electronics Engineers Transactions on Instrumentation and Measurement",
    issn = "0018-9456",
    publisher = "IEEE, Institute of Electrical and Electronics Engineers",
    number = "3",

    }

    TY - JOUR

    T1 - An Integrated Framework for 3-D Modeling, Object Detection, and Pose Estimation From Point-Clouds

    AU - Guo, Yulan

    AU - Bennamoun, Mohammed

    AU - Sohel, Ferdous

    AU - Lu, M.

    AU - Wan, J.

    PY - 2015/3

    Y1 - 2015/3

    N2 - 3-D modeling, object detection, and pose estimation are three of the most challenging tasks in the area of 3-D computer vision. This paper presents a novel algorithm to perform these tasks simultaneously from unordered point-clouds. Given a set of input point-clouds in the presence of clutter and occlusion, an initial model is first constructed by performing pair-wise registration between any two point-clouds. The resulting model is then updated from the remaining point-clouds using a novel model growing technique. Once the final model is reconstructed, the instances of the object are detected and the poses of its instances in the scenes are estimated. This algorithm is automatic, model free, and does not rely on any prior information about the objects in the scene. The algorithm was comprehensively tested on the University of Western Australia data set. Experimental results show that our algorithm achieved accurate modeling, detection, and pose estimation performance.

    AB - 3-D modeling, object detection, and pose estimation are three of the most challenging tasks in the area of 3-D computer vision. This paper presents a novel algorithm to perform these tasks simultaneously from unordered point-clouds. Given a set of input point-clouds in the presence of clutter and occlusion, an initial model is first constructed by performing pair-wise registration between any two point-clouds. The resulting model is then updated from the remaining point-clouds using a novel model growing technique. Once the final model is reconstructed, the instances of the object are detected and the poses of its instances in the scenes are estimated. This algorithm is automatic, model free, and does not rely on any prior information about the objects in the scene. The algorithm was comprehensively tested on the University of Western Australia data set. Experimental results show that our algorithm achieved accurate modeling, detection, and pose estimation performance.

    U2 - 10.1109/TIM.2014.2358131

    DO - 10.1109/TIM.2014.2358131

    M3 - Article

    VL - 64

    SP - 683

    EP - 693

    JO - Institute of Electrical and Electronics Engineers Transactions on Instrumentation and Measurement

    JF - Institute of Electrical and Electronics Engineers Transactions on Instrumentation and Measurement

    SN - 0018-9456

    IS - 3

    ER -