Plausibility-based approach to eliminate line pattern indexing ambiguities

Single-shot multi-line triangulation, a common 3D measurement principle, suffers from line indexing ambiguities. Ambiguity occurs when parts of the projected pattern are missing. With a proper geometry it is possible to avoid ambiguities, but only inside a restricted measurement volume (S. Ettl et al., Appl Opt 51, 2012). More information has to be acquired to eliminate ambiguities. One option is to add one or more cameras (F. Willomitzer et al., DGaO 2013 & 2014). Another option is to exploit spatial and/or temporal context information. We present a novel indexing method which explores plausibility information. The basic idea is to use a line pattern for which each nth line (n>>1) is encoded. These few lines with known index are used to generate a sparse 3D model of the object. Based on this model, the remaining lines are successively indexed by performing an a posteriori plausibility check. As an important consequence, the depth of the measurement volume can be extended and the number of projected lines can be increased, as a further step towards a better single-shot 3D camera. Experimental results will be presented to demonstrate the potential of this novel method.