Geometrically Optimized PTZ Camera Calibration From Two Images

We make three main contributions to autocalibration under pure rotation and zooming, with particular focus on PTZ cameras. First, for the general camera rotation, we provide a novel solution that yields five intrinsic parameters from only two images, by directly decomposing the infinite homography using a series of Givens rotations. Second, we present thorough analysis of degenerate cases of pure pan and pure tilt that occur very frequently in practical applications of PTZ cameras, and hence reveal the limits and the stability of auto-calibration under these degenerate rotations. Third, we derive a new optimal geometric error function for pure rotation, which plays a similar role as the epipolar constraint in a freely moving camera, in terms of characterizing the reprojection error of point correspondences. Our solutions and analyses are thoroughly validated and tested on both synthetic and real data, whereby the new geometric error function is shown to outperform existing methods in terms of accuracy and noise resilience.