A calibration trilogy of monocular-vision-based aircraft boresight system

Aligning the orientation of Mounting Bases (MB) of sensors/weapons relative to the aircraft's reference datum, namely aircraft boresighting, is an important but tough task in aircraft assembly. A prototype MOnocular-Vision-based Aircraft Boresight (MOVAB) System, including a CCD camera and several Boresight Units (BU), is developed. The relative orientations of the MBs are obtained by capturing only one image of the BUs, which are mounted on the MBs. A BU is devised including a Visual Target Panels (VTP) with Infrared Light Emitting Diodes (ILED) as targets, an adapter which represents the MB and a bracket which bridges the VTP and the adapter. Since the camera of the MOVAB has a large working volume (i.e. large field of view and large depth of field), it is hard to make a calibration object large enough to guarantee the camera calibration accuracy. To this end, a new camera calibration method is proposed, which is based on the images taken from multiple perspectives of a 3D virtual calibration target generated with the aid of a coordinate measuring machine. The exact positions of the ILEDs on the VTP are calibrated based on the multiple view geometry principle. To recover the absolute metric of the layout of the target points, two auxiliary target points with known separation are constructed by means of accurate movement and are incorporated in the photogrammetry process. A concise yet effective procedure is proposed to calculate the relative attitude between the VTP and the adapter. It only needs to move the BU along two perpendicular lines and take images. The trio calibration is experimentally validated, and the aligning result from the MOVAB is compared to that a high-precision laser tracker.