Ada3D: Real-Time 3D Reconstruction with Compression Rate Adaptation.

2020 
Augmented reality, so called AR, is recently the subject of attention by researchers from different fields and is developing rapidly. However, some attractive AR applications do not have mature technical basis yet, such as holographic conference and telesurgery. In these applications, 3D reconstruction of the target objects and real-time transmission of the reconstructed result are needed. It is challenging to build an integrated real-time 3D reconstruction system and effectively reduce the transmission delay within the system. In this paper, we propose Ada3D, a practical cross-platform system for real-time 3D reconstruction using multiple RGB-D sensors. It consists of several sensory clients and an edge server, which are responsible for data acquisition and data fusion, respectively. In addition, a compression-based method is applied in the system, which enables the client to automatically select an appropriate data compression level according to the current system state, including network status and device capability. Experiment results show that the method outperforms the common case by over 10% in delay saving.1
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    4
    References
    0
    Citations
    NaN
    KQI
    []