Gradient and Curvature from Photometric Stereo Including Local Condence Estimation
Robert J. Woodham · 1994
Photometric stereo is one technique for 3D shape determination that has been implemented in a variety of experimental settings and that has produced consistently good results. The idea is to use intensity values recorded from multiple images obtained from the same viewpoint but under dieren t conditions of illumination. The resulting radiometric constraint makes it possible to obtain local estimates of both surface orientation and surface curvature without requiring global smoothness assumptions and without requiring prior image segmentation. This paper moves photometric stereo one step closer to practical viability by describing an experimental setting in which surface gradient estimation is achieved on full frame video data at near video frame rates (i.e., 15Hz). The implementation uses commercially available hardware. Reectance is modeled empirically using measurements obtained from a calibration sphere. Estimation of the gradient, (p; q), requires only simple table lookup. Curvature estimation uses, in addition, the reectance map, R(p; q). The required lookup table and reectance maps are derived during calibration. Because reectance is modeled empirically, no prior physical model of the reectance characteristics of the objects to be analyzed is assumed. At the same time, if a good physical model is available, it can be retrot to the method for implementation purposes. Photometric stereo is subject to error in the presence of cast shadows and interreection. No purely local technique can succeed since these phenomena are inherently non-local. Nevertheless, this paper demonstrates that one can exploit the redundancy in three light source photometric stereo to, in most cases, locally detect the presence of cast shadows and interreection. Detection is facilitated by explicitly including a local condence estimate in the lookup table used for gradient estimation.