Platform calibration of an omnidirectional vision system composed of a camera and a conical mirror

Abstract

Omnidirectional vision systems allow for a 360° imaging field and are applied in many areas of knowledge, including close range Photogrammetry. Obtaining reliable measurements, the Photogrammetry's goal, requires calibration of the system, a step in which the parameters of the mathematical models are estimated. The calibration of the system is divided into internal calibration and platform calibration. The main objective of this work is to perform the calibration of the platform of an omnidirectional vision system composed of a camera and a conical mirror. In the calibration of the platform we estimated the parameters that relate the imaging system to the georeferencing system, composed of a GNSS (Global Navigation Satellite System) receiver and an IMU (Inertial Measurement Unit). In the calibration of the platform the boresight angles of the IMU were estimated related to the cone reference and the lever arm of the vertex of the cone to the center of the GNSS antenna. The uncertainty in the estimation of the lever arm is at least 10 times less than the precision obtained with the GNSS receiver in kinematic mode, which indicates an adequate quality for the estimated value. Regarding the misalignment angles, the RMSE (Root Mean Square Error) was higher for the rotation angle kIMUCone, around 30 minutes of arc, and for the other angles, the RMSE was around 10 minutes of arc.