Lambertian Reflectance and Linear Subspaces-英文文献.pdf

Weizmann Institute of Science Technical Report MCS NEC Research Institute Technical Report R Lamb ertian Reectance and Linear Subspaces Ronen Basri David Jacobs Dept of Computer Science NEC Research Institute The Weizmann Institute of Science Indep endence Way Rehovot Israel Princeton NJ Abstract We pr ove that the set of al l reectance functions the mapping fr om surface normals to intensities pr oduced by Lambertian objects under distant isotropic lighting lies close to a D linear subspace This implies that in general the set of images of a convex Lambertian object obtained under a wide variety of lighting conditions can be approximated accurately by a lowdimensional linear subspace explaining prior empirical results We also pr ovide a simple analytic characterization of this linear space We obtain these results by representing lighting using spherical harmonics and describing the eects of Lambertian materials as the analog of a convolution These results al low us to construct algorithms for object recognition based on linear methods as wel l as algorithms that use convex optimization to enforce nonnegative lighting functions Final ly we show a simple way to enforce nonnegative lighting when the images of an object lie near a D linear space  Research conducted while at NEC Research Institute Princeton NJ Ronen Basri is an incumb ent of Arye Dissentshik Career Development Chair at the Weizmann Institute Intro duction One of the most basic problems in vision is to understand how variability in lighting aects the i