Extracting Surface Curvature from Noisy Scan Data
John Rugis · 2006
In general, the noise that is present in real-world 3D surface scan data prevents accurate curvature calculation. In this paper we show how curvature can be extracted from noisy data by applying filtering after a noisy curvature calculation. To this end, we extend the standard Gaussian filter (as used in 2D image processing) by taking adjacent point distances along the scanned surface into account. A brief comparison is made between this new 2.5D Gaussian filter and a standard 2D Gaussian filter using data from the Digital Michelangelo Project.