Shape from interreflections

作者:Shree K. Nayar, Katsushi Ikeuchi, Takeo Kanade

摘要

Shape-from-intensity methods assume that points in a scene are only illuminated by the sources of light. This assumption is valid only when the scene consists of a single convex surface. Most scenes consist of concave surfaces where points reflect light among themselves. In the presence of these interreflections, shape-from-intensity methods produce erroneous (pseudo) estimates of shape and reflectance. This article shows that, for Lambertian surfaces, the pseudo shape and reflectance are unique and can be mathematically related to the actual shape and reflectance of the surface. We present an iterative algorithm that simultaneously recovers the actual shape and reflectance from the pseudo estimates. The general behavior of the algorithm and its convergence properties are discussed. Simulations as well as experimental results are included to demonstrate the accuracy and robustness of the algorithm.

论文关键词:Image Processing, Artificial Intelligence, Computer Vision, Computer Image, Iterative Algorithm

论文评审过程:

论文官网地址:https://doi.org/10.1007/BF00115695