Study on Typical Objects Three-Dimensional Modeling and Classification Technology Based on UAV Image Sequence

2021 
The cost of data acquisition is relatively high in airborne LiDAR platform, not suitable for the entire industry promotion. Point-Cloud extraction technique based on image, a method based on low-altitude drone image and flight control data for 3D reconstruction is adopted. Do not depend on prior knowledge of camera calibration or other premise, multi-angle, high-overlap image sequence acquired by visible light camera, to restore the three-dimensional reconstruction information. Aiming at the problem of insufficient memory when dealing with high-resolution unmanned aerial vehicle (UAV) images by SIFT (Scale-Invariant Feature Transform) feature extraction operator, an improved SFIT algorithm for image segmentation is proposed. The point cloud filtering is carried out by the method of orthogonal polynomial banding filtering, the ground point and the non-ground point are separated. The non-ground point is used to extract the roof set of the building. This experiment has obtained good test results based on the UAV sequence image in the test area, which has certain application value for low-cost building 3D modeling, 3D modeling of digital city, and damage building damage identification.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    15
    References
    0
    Citations
    NaN
    KQI
    []