Non-Parametric Structure-Based Calibration of Radially Symmetric Cameras Supplementary Material

In this supplementary material, we present additional re-sults of the experiments shown in the main paper. As men-tioned in the original submission, the method for calibration is very general and can handle a wide array of radially sym-metric cameras. To emphasize this, in the paper we showed results for the more extreme of these distortions: three cata-dioptric cameras (including a non-central one) and one fish-eye camera. In this supplementary material, however, we want to demonstrate the applicability of the method to other very common camera lens types, such as a pinhole cam-era and popular Wide Field of View (WFOV) cameras. In addition to the model used in the paper, we also present cal-ibrations obtained using the Aachen model from [2]. The experimental results for these additional camera types are presented similarly to the ones detailed in the pa-per, i.e., by showing the calibration mapping as a scatter plot and comparing the data points to well-known calibra-tion toolboxes [1, 3]. As stated in the paper, our calibra-tions are obtained by utilizing the geometric ordering con-straint. The scatter plots of Fig. 2 are color-coded accord-ing to the image that generated the matches, since for all calibration multiple images were needed. This thus illus-trates the amount of images used and the concentration of the scatter near the reference calibration. Such reference calibrations are shown as thick line plots in gray behind the scatter plot. In Fig. 2a, Fig. 2b, and Fig. 2c we show calibrations ob-tained for the Sony ExmorR, GoPro3, and an iPhone 4S (no lens attachment), respectively. In the case of Fig. 2a, we can see that its calibration very closely matches the one ob-tained using [3]. For high radii, however, our calibration appears more accurate (c.f. Fig. 1). Additionally, to quali-tatively demonstrate the quality of the calibrations obtained with our method, we present undistorted images for several of the cameras used in the paper