A Toolbox for Easily Calibrating Omnidirectional Cameras

摘要: In this paper, we present a novel technique for calibrating central omnidirectional cameras. The proposed procedure is very fast and completely automatic, as the user only asked to collect few images of checker board, click on its corner points. contrast with previous approaches, does not use any specific model sensor. It assumes that imaging function can be described by Taylor series expansion whose coefficients are estimated solving four-step least-squares linear minimization problem, followed non-linear refinement based maximum likelihood criterion. To validate technique, evaluate performance, apply calibration both simulated real data. Moreover, show accuracy projecting color information calibrated camera 3D points extracted sick laser range finder. Finally, provide Toolbox which implements procedure.