Objectives  To satisfy the widely demand of 3D real scene reconstruction, a novel unmanned aerial vehicle (UAV) photogrammetric technique approach, named optimized views photogrammetry, is presented.

  Methods  In terms of the 3D rough model of the scene which obtained in various ways, optimized views photogrammetry generates initial views via stereoscopic observation sampling and selects optimal views under the constraint of observability measure, and then forms the aerial photogrammetry planning path for quadcopters. For the contradiction between large-scale application and limited endurance of UAV system, optimized views photogrammetry provides a multi-UAV collaborative survey solution with its specific path division function in the planning end. A practical application verification was performed in urban central district of Qingdao, to assess the performance of multi-UAV cooperative optimized views photogrammetry in relatively large range complicated urban scene.

  Results  The output results of real 3D reconstruction postprocessing and the detailed comparison with the model generated from oblique photogrammetry verify that, on the premise of high-quality real scene 3D output, the technical approach realizes the real scene 3D acquisition in a large range of complex space with light and small aerial photography equipment, as meanwhile and improves the efficiency through collaborative operation.

  Conclusions  With the key support of optimized views photogrammetry, it could also make an extension to the new technology application mode of ubiquitous real 3D data acquisition for multiple technical directions and application fields.