Tomographic Triangulation

We will present a novel method for a single-shot optical 3D-camera. The motivation justifies the challenge: A single-shot 3D-camera would enable the acquisition of 3D-movies even of objects in motion. Information-theoretical considerations [G. Häusler, Proc. of the DGaO 2012] reveal that a (monomodal, high resolution) 3D-camera requires more than one set of raw images for each single 3D-image. Three images might be sufficient, but noise and better lateral resolution suggest using more images. A paradigm of a “multi-image-concept” is X-ray CT. We have adapted the concept of X-ray CT to the optical domain: 1. We simultaneously take images from many directions, by a set of cameras. 2. In order to get corresponding points at any kind of surface, structured illumination is applied. 3. The 3D-reconstruction is done via back-propagation of the raw images. 4. An a posteriori filtering is necessary, like in X-ray CT - which is, in the optical domain, implemented via a Hough concept. Experiments display the proof of the concept.