The Development of a Skin Chart Based on Real Measured Data

In this paper, a methodology is proposed for uniformly selecting a specified number (N) of colors for a skin chart from the color solid in CIELAB space under D65/2° viewing condition, based on the ISSA database. The method generates two distinct sets of colors. One set is formed by the centers of N equal-sized cubes, which cover the whole color solid in CIELAB space formed by the ISSA dataset, with each cube containing at least one skin data from the ISSA database. Hence the centers represent skin colors that are uniformly distributed within and around the color solid defined by the ISSA database. Furthermore, reflectance is reconstructed for each of the centers. Another set of N colors are the real skin colors from the ISSA database that are nearest to each of the N centers. Comparison results show both sets of colors selected by the proposed method outperform the colors selected by the k-means and PCA/SVD method using the ISSA database, in terms of color volume and uniformity of distribution within the color solid. Hence, the skin chart developed from colors selected by the proposed method represents the global skin color diversity, and offers potential applications in cosmetics, fashion, printing, photography, and computer vision.