Geometric calibration of rotational kaleidoscopic instrument

Abstract

The measurement of spatially varying surface reflectance is required for faithful reproduction of real world to allow for predictive look of computer generated images. One such proposed method uses a rotational kaleidoscopic imaging, where illumination and imaging paths are realized by subimages on kaleidoscopic mirrors and illumination is carried out by a DLP projector. We describe a novel geometric calibration method for a rotational kaleidoscope that is necessary to get aligned and accurate data from measurement. The calibration has two stages. The first stage mechanically adjusts the camera, the projector, and the autocollimator against the kaleidoscope mirrors. The second stage is based on the software. By random perturbation of camera and projector in corresponding mathematical model of the kaleidoscope we estimate better real positions of camera and projector in a physical setup, comparing the computed images from the software simulator and the acquired images from the physical setup.