This thesis presents a new approach to automatically acquire accurate high-resolution range images of close-range scenes in real-time. The technique combines and extends the principles of the color-coded light approach and stereo-vision. Moreover, it develops solutions to related problems such as the calibration of the acquisition system, i.e. to camera and projector calibration. A prototype system based on the presented ranging approach, integrated using low-cost off-the-shelf components, is evaluated. Experiments show that it is able to acquire range maps of resolution 780 by 580 at up to 25 frames per second with an exemplary measurement accuracy of 0.2 mm standard deviation over a cubical working space of about 0.5 m side length. Finally, this thesis shows that the performance of a face recognition system can be significantly improved by the use of range data as acquired by the prototype.
«
This thesis presents a new approach to automatically acquire accurate high-resolution range images of close-range scenes in real-time. The technique combines and extends the principles of the color-coded light approach and stereo-vision. Moreover, it develops solutions to related problems such as the calibration of the acquisition system, i.e. to camera and projector calibration. A prototype system based on the presented ranging approach, integrated using low-cost off-the-shelf components, is ev...
»