Speaker: Daniel Prieler (ICGA)
The increasing availability of 3D scanning devices in both industrial and entertainment environments (e.g. Kinect) creates a demand for fast, reliable resampling and reconstruction techniques. Point clouds, especially raw range images, are often non-uniformly sampled and are subject to non-uniform noise levels. We present an iterative point cloud resampling method that estimates and adapts to the local noise level at each sample. Both the resampling algorithm and the subsequent consistent normal orientation operate locally for efficient parallel implementation. The iterative algorithm permits recovering anisotropic neighborhoods to account for the non-uniform sampling which in turn provides more accurate surface reconstruction.