Simultaneous calibration of a colour camera, depth camera, and projector and 3D object pose recognition

Laura Luidolt

Supervisor(s): Andreas Reichinger

VRVis Research Center

Abstract: In this paper, we present a novel approach to calibrate a 4-component setup, consisting of a movable object beneath a static arrangement of a colour camera, a depth camera, and a projector, which was developed as part of a tactile audio guide to help visually impaired people perceive images. For the static arrangement, we propose a procedure that allows calibrating all three devices simultaneously using a single calibration target with two different calibration patterns – one printed and one projected. The application is easy to use, due to the mostly automated calibration process and the algorithm still achieves reasonable accuracy for the devices and their respective poses. For the movable object, a point cloud matching algorithm is implemented to detect its 3D pose with respect to the three device coordinate systems, when placed beneath the setup. This allows to register interaction gestures to specific areas on the object and to project images onto the object.
Keywords: Computer Vision, Image Processing, Point-based Graphics
Full text:
Year: 2018