作者: Jared Heinly , Enrique Dunn , Jan-Michael Frahm
DOI: 10.1007/978-3-319-10593-2_51
关键词:
摘要: Structure from motion (SfM) is a common technique to recover 3D geometry and camera poses sets of images scene. In many urban environments, however, there are symmetric, repetitive, or duplicate structures that pose challenges for SfM pipelines. The result these ambiguous incorrectly placed cameras points within the reconstruction. this paper, we present post-processing method can not only detect errors, but successfully resolve them. Our novel approach proposes strong informative measure conflicting observations, demonstrate it robust large variety scenes.