In recent years,3D reconstruction technology based on UAV tilted photographic images has been a research hotspot in the field of photogrammetry and computer vision,and has been widely used in the production of large-scale topo-graphic maps as well as in the field of smart cities.In view of the problems of high data redundancy and high time and space complexity of UAV tilted images,this paper proposes a fast and effective 3D reconstruction method for UAV sequence images.Sequential UAV images of a section of the open scene of the Middle Route of the South to North Water Transfer Project were selected to delineate the image zones by relying only on the matching results among the overlapping images without assisting other localization information such as POS.The improved incremental SFM method is used for 3D recon-struction within individual partitions,and finally each partition is merged based on the overlapping information between par-titions to complete the 3D reconstruction of the entire survey area.It is compared with commercial modeling software and traditional computer vision 3D reconstruction methods.The results show that the method can significantly improve the re-construction efficiency while maintaining high accuracy,and the reconstruction speed can be improved by more than 5 times,which is especially suitable for large-scale datasets with large scenes.
维普
万方数据