Optimized views photogrammetry and its precision analysis

UAV path planning and accurate data acquisition is a key problem for 3D reconstruction of complex urban scenes. By using the rough model of the object to be captured, optimized views photogrammetry combines the technique of dense sampling-based initial view generation and reconstructability constrai...
Ausführliche Beschreibung

UAV path planning and accurate data acquisition is a key problem for 3D reconstruction of complex urban scenes. By using the rough model of the object to be captured, optimized views photogrammetry combines the technique of dense sampling-based initial view generation and reconstructability constrained view optimization and achieves the purpose of precious 3D model reconstruction from the least number of images. First, this study presents the principle of optimized views photogrammetry. Second, data acquisition campaigns of real scenes are conducted with the aid of optimized views photogrammetry. Combining laser scanned point clouds, the assessment of 3D reconstruction precision and mesh model quality is performed. Experimental results verifies that for the building facades that are hardly observed from oblique photogrammetry views, the mesh model quality is improved dramatically with the increase ratio of 3 to 5 times. In addition, compared with traditional oblique photogrammetry, optimized views photogrammetry can establish stable image connection network and achieve competitive precision in absolute bundle adjustment. Ausführliche Beschreibung