Image measurement and interpretation based on goniometrical textural data

Abstract

Sparkle is a visual texture which appearance basically consists of very bright spots randomly distributed over a darker background. The effect is more apparent to the observer when one of the contributors (sample, observer, illumination) is moved, meaning it is a dynamic effect. While the characterization and measurement of sparkle has been an important topic of research during the past years, the focus has been solely around the static component of it, the dynamic aspect of sparkle has been left out of the discussion. This is somewhat reasonable since the dynamic attribute bases itself on the perception and measurement of static sparkle. The dynamic aspect of sparkle, or “living sparkle” can be described as the variation of its appearance over a given angular distance. Nonetheless, both effects are without doubt related and perhaps together they provide the key to finding the absolute characterization of sparkle. In this work we aim to dive into both aspects. First, the setup of a goniometric system with a color camera was carried out, obtaining a high spatial and angular resolution. Second, measurements of commercial and self-manufactured sample sets were performed, and an analysis algorithm was developed. The algorithm provided a wide range of parameters for different thresholding methods. These methods were put to test when the results were correlated with the visual assessment of a group of 9 samples under one illumination/observation geometry, obtaining the correlation factors for all methods and parameters. The combination of thresholding method and parameter obtained an R2 > 0.9, implying a strong connection with the visual experience. The threshold applied was always fixed at the same value and the parameter (“S_Values”) represented the brightness of the pixels, indicating that the brightness of the sample is the most influential attribute for the perception of sparkle. Then, this same methodology was applied to the rest of the samples in this study (71 total), for the geometries in between 71.2 and 11.2 degrees in 5 degrees steps. The physical parameters of the self-manufactured samples were also used to test the correctness of the algorithm. The results broadly matched the expected values of intensity and particle size. In addition, the algorithm measured the angular variability of the samples though the angular visibility of the sparkle spots. The wider the visible range, the more invariant the sample, and the less apparent the sparkle effect. This work proposes an innovative way of measuring sparkle in its dynamic and static components, a methodology and measurands to do so are proposed and contrasted to visually assessed samples.