Modeling, measurement, and calibration of three-axis integrated aerial camera pointing errors

Abstract

Aerial cameras are currently widely used in various fields. However, traditional aerial cameras have certain limitations. We propose a new three-axis integrated aerial camera that can significantly improve imaging efficiency. Both traditional aerial cameras and three-axis integrated aerial cameras suffer from reductions in image quality due to mechanical errors. In this paper, we analyze the influence of mechanical errors on pointing errors and establish a pointing parametric model (PM) based on spatial coordinate transformation and ray tracing. We also consider how the presence of nonlinear errors in aerial cameras affects image quality, and propose a semi-parametric model (SPM) to compensate for nonlinear errors. The PM and the SPM can be used to calibrate the pointing of aerial cameras. In addition, we propose an improved measurement method that can gauge the pointing errors that occur when the three-axis integrated aerial camera rotates to different angles. The results prove that both the PM and the SPM can effectively calibrate the pointing of the aerial camera. Following calibration using the SPM, the pointing errors were greatly reduced. The mean was reduced by more than two orders of magnitude, and the variance was reduced by 99.95%. The SPM completes the pointing calibration better than the PM.

Pointing Error Correction for Vehicle-Mounted Single-Photon Ranging Theodolite Using a Piecewise Linear Regression Model

“…Xu et al utilized the wavelet denoising algorithm based on a soft threshold to estimate the nonlinear error and correct pointing error of the airborne photoelectric platform [15]. Liu and Huang, respectively, employed the compensated least square method to estimate and correct the nonlinear errors of aerial cameras and telescopes [16,17]. Additionally, Peng et al applied the K-Nearest Neighbor Algorithm and kernel weight function to achieve successful correction of pointing errors in optical communication terminals [18,19].…”

Metrology and Measurement Systems

“…A robust approach using local multi-cameras and non-linear optimization has also been proposed [22,23] but the number of cameras may increase the false matches and decrease the S/N ratio [24]. Other methods such as modelling the motion blur [25], the back projection process [26], trained neural network [27], rotating axis calibration [28,29] or pointing error methods [30] can also model the uncertainty and improve the vision-based system accuracy.…”