3D cameras: Errors, calibration and orientation

Abstract

Range cameras integrate different optical measurement techniques that provide coverage of an area. Firstly, they provide images like frame cameras which provide scene information in the form of texture. Secondly, they provide direct range measurement like laser scanners, but do so simultaneously for the entire field of view as opposed to the sequential operation of laser scanners. Above all, they provide image streams like video cameras. For specific measurement and modeling tasks, one observation technology, i.e. either passive imaging or laser scanning, is typically more suited than the other one. The integrated aspects of 3D cameras have therefore triggered a lot of interest despite their relative low resolution and accuracy. The potential to obtain 3D scene information instantly and directly thus caused different groups to investigate the data quality of 3D range cameras, models for calibration and orientation.