BRDF valid sampling based on gradient magnitude synthetic analysis

In general, there are some minor flaws on the surface of homogeneous material, such as stains, nicks, will produce unfavorable effects on isotropic BRDF measurement, especially for highlight material. In this paper, we present a numerical technique, which can efficient solve this problem by gradient magnitude synthetic analyse. We measure BRDF by fixing camera and rotating light source. We found that nicks or stains will exhibit different properties from their surrounding material in some special light orientations. A novel algorithm is defeloped to detect tiny nicks or stains based on this principle. We merge gradient magnitude info of possible edge points by common edge detection algorithms at all light source orientations and classify edge points as several ranks, and then check invalidity of edge points and rational distribute them by comprehensive analyze the rank of each point.