地面视频流与无人机影像融合建模技术OA

This paper addressed the challenges encountered in traditional unmanned aerial vehicle(UAV)-based oblique pho-togrammetry,particularly in obtaining images of sufficient precision and clarity for the sides and bottom of structures.Using a sculpture as a case study,the paper proposed a method that combined multi-view images from aerial UAVs with continuous video streams captured from the ground.This approach involved extracting significantly reliable continuous frames from the ground video stream and registering these images based on the spatial feature points from the UAV's multi-view image model.Subsequently,the results of the aerotriangulation from both aerial and ground data sources were fused to achieve model construction.The paper detailed the steps involved in the fusion modeling framework,including data preprocessing,feature extraction,fusion algorithms,and model construction through a practical example.Experimental results show that the mean distance error and mean height error of the fusion model are both less than 1 cm.This method not only enriches tra-ditional data collection techniques for image modeling,achieving a complementary advantage of different modeling approaches,but also avoids the issue of insufficient precision in expressing local details in traditional UAV photogrammetry modeling.