A survey on extrinsic self-calibration algorithms for cameras without overlapping field of views

Extrinsic self-calibration of cameras, i.e., to automatically determine the extrinsic parameters of the cameras (position and orientation), is regarded as an important preparatory requirement. However, if the cameras do not share overlapping field of views (FoVs), they lack the common information, which is normally obtained from the shared FoVs and used to design the self-calibration algorithms. Hence, it is more difficult and challenging to design the extrinsic self-calibration algorithm in non-overlapping camera networks. In this paper, the existing extrinsic camera self-calibration algorithms for non-overlapping camera networks are reviewed, classified and summarized in detail.