Shape from single stripe pattern illumination

Abstract

This paper presents a strategy for rapid reconstruction of surfaces in 3d which only uses a single camera shot of an object illuminated with a simple stripe pattern. With this respect, it is a meaningful extension of our 'shape from 2d edge gradient' method introduced earlier. The reconstruction is based on determining stripe directions and stripe widths in the camera image in order to estimate surface orientation. I.e., this method does not use triangulation for range data acquisition, but rather computes surface normals. These normals can be 2d integrated and thus yield the surface coordinates; in addition they can be used to compute robust 3d features of free-form surfaces for object recognition, pose estimation, etc. The method is straightforward and very efficient by processing only one image and using only simple image processing operations. © Springer-Verlag Berlin Heidelberg 2002.