Estimating reflectance parameters, light direction, and shape from a single multispectral image

This paper presents a novel approach for estimating the light direction, shape, and reflectance parameters from a single multispectral image. We start from a general formulation that hinges in the notion that the light reflected from an object can be deemed to be a linear combination of specular and diffuse reflections. This permits the recovery of the reflection parameters through an iterative optimization scheme, which we render well posed by adopting a novel reparameterization that reduces the number of degrees of freedom in the cost function. With the estimated specular reflectance parameters, we recover the single point light source position from specular highlights by applying two novel constraints, coplanarity and Kullback-Leibler divergence. Then, by integrating the knowledge of light source and diffuse reflectance parameters, we recover shape of the scene from the diffuse component. Our approach is quite general in nature and can be applied to a family of reflectance models that are based on the Fresnel reflection theory. We demonstrate the utility of our method on synthetic and real world imagery. We also compare our results to several alternatives in the literature.