Measuring and Rendering Art Paintings Using an RGB Camera

Abstract

A method is proposed for measuring and rendering of art paints by using the simple system of an RGB camera only. The surface shape of an art painting is considered as a rough plane rather than a three-dimensional curved surface. Because the surface material has the dichromatic reflection property, the spectral reflectance function is estimated from the diffuse reflection component. First, we present an algorithm for estimating surface normal at each pixel point, based on an extended photometric stereo without using a rangefinder. Next, an algorithm is presented for estimating the spectral reflectance function from a set of RGB color values acquired at different illumination directions. Then, the surface reflectance and normal data are used for estimating the light reflection properties. The Torrance-Sparrow model is used for the model fitting and parameter estimation. Finally, an experiment using an oil painting is executed for demonstrating the feasibility of the proposed method.