Calibration and registration framework for multisensor panoramic color scanning

Abstract The focus of this work is 'Modeling the Reality'. The goal is the reconstruction of scenes with accurate geometry and real radiometric textures. Thereby, a common problem is to acquire colored panoramic 3D data, which is highly accurate with sufficiently high resolution. We present a novel, multi-sensor calibration and registration framework that fuses highly accurate and robust colored panoramic 3D data. To achieve this, we use a panoramic 2.5D laser scanner in combination with a 2D matrix camera and an electronic spirit-level. The spatial resolution of the scanner is very dense and requires high precision for sensor calibration and for the registration of data streams. We introduce a novel sensor model for the scanner and describe it in a general context for calibrating polar measurement systems. For covering the same visual field for the digital camera, it is mounted on a vertical tilt-unit. Overlapping color images are acquired by rotating the entire system horizontally and the tiltunit vertically by predefined angle increments. The calibration of the camera and the registration between both sensors can be enormously simplified by using these increments as fixed input parameters.