A Numerical Analysis of a Frame Calibration Method for Video-based All-Sky Camera Systems

The field of meteor monitoring has grown considerably over the past 20 years with the development of affordable, automated video camera systems. We describe a method for calibrating video all-sky cameras in terms of local zenith and azimuth angles. The method involves the observation of known training points (stars) and is based on an approach developed by Ceplecha & Borovička. We use a simplified equation set, incorporate a quadratic expression for modeling the lens response, and utilize a nonlinear solver to obtain the calibration parameters. Simulation results with synthetic star data are presented to examine the effect of a limited number of training points, training point location, and initial parameter values on the calibration. Assumed simulation parameters are consistent with expectations for cameras in the NMSU SkySentinel network. Our modified calibration approach is shown to be stable over a broad range of calibration parameters with typical azimuth and zenith residual errors of less than 1°. Example calibration results for three camera nodes in the SkySentinel network are presented.