Communications and Networking. 12th International Conference, ChinaCom 2017, Xi’an, China, October 10-12, 2017, Proceedings, Part II

Research Article

Combining Computer Graphics and Feature for 3D Camera Tracking Based on CAD Model

  • @INPROCEEDINGS{10.1007/978-3-319-78139-6_3,
        author={Linlin Wang and Guoyun Lv and Ningxin Zhang and Yanggege Yu},
        title={Combining Computer Graphics and Feature for 3D Camera Tracking Based on CAD Model},
        proceedings={Communications and Networking. 12th International Conference, ChinaCom 2017, Xi’an, China, October 10-12, 2017, Proceedings, Part II},
        proceedings_a={CHINACOM},
        year={2018},
        month={4},
        keywords={Template-based tracking Feature-based tracking CAD model},
        doi={10.1007/978-3-319-78139-6_3}
    }
    
  • Linlin Wang
    Guoyun Lv
    Ningxin Zhang
    Yanggege Yu
    Year: 2018
    Combining Computer Graphics and Feature for 3D Camera Tracking Based on CAD Model
    CHINACOM
    Springer
    DOI: 10.1007/978-3-319-78139-6_3
Linlin Wang1,*, Guoyun Lv1,*, Ningxin Zhang1,*, Yanggege Yu1,*
  • 1: Northwestern Polytechnical University
*Contact email: 534183837@qq.com, 178510887@qq.com, 2568117320@qq.com, 376151248@qq.com

Abstract

A real-time 3D tracking system based on CAD model is proposed in this paper. The strategy in the tracking process includes both template-based and keypoint-based approaches. Compared with traditional CAD model-based tracking, a more accurate initial camera pose can be provided for the following feature-based operation which can accelerate the convergence of camera pose estimation. Using template-based method reduces the demand for textures of the tracking object to improve the universality of the system. Furthermore, adaptive visual feature extraction within the feature-based tracking is adopted in the experiment, and feature homogenization and other methods is also chosen to enhance the robustness and interference immunity of the system. Finally, the effectiveness and stability of the methods proposed in this paper is verified through the experiment of tracking targets in the image sequence.