Method for estimating scene parameters from color histograms

This paper presents the characterization of a color histogram of a dielectric object by a few measurements directly related to several scene properties. It is shown that surface roughness, imaging geometry, illumination intensity and illumination color can be extracted from these histogram measurements, which can be used to calculate the object color and albedo. The algorithm developed, which is found to be independent of the shape of the object, can be applied to real images for reasonable accurate estimates of phase angle, illumination intensity, and surface roughness. Results from test of the algorithm on simulated data and of the technique on real images gives good results. The advantages of the method are its speed and the fact that only a single color image is required.