서강대학교

초록/요약

Since the advent of the wide gamut displays, such as LED-backlit and Laser, there have been many literatures introducing gamut extension algorithms (GEAs) to take full advantages of the wide gamut displays. Most of the gamut extension algorithms are focused on the preservation of the hue of the source image or the minimization of hue shifts while scaling the saturation of the source image and modifying its lightness because in the scientific points of view, the preservation of the hue of the source gamut is essential, when it expands to the reproduced gamut. From artistic points of view, hue shifts can be regarded as artists’ preference in gamut extension algorithms. Dubail et al. mention about color grading artists’ common behavior in color transformation; color grading artists tend to increase saturation to reach the maximum saturation available on devices and move the hues toward primary colors, R, G, and B to get colors with more purity [5]. Our fundamental goal of this study is to develop a new GEA which can hold both scientific and artistic perspectives; our GEA considers hue as one of the color attributes to modify with saturation. Chiefly, our GEA is divided into two parts: a hue mapping function and a saturation mapping function. As the hue mapping function, we use a curve extension method, and as the saturation mapping function, we use three asymmetric Gaussian functions. Our GEA consists of a number of curves that expand toward each RGB primary of the destination gamut, and the three asymmetric Gaussian functions which can give different saturation intensity to each RGB. Since our GEA is composed of curves expanding toward each RGB primary of the destination gamut, hue shifts toward each RGB primary color inherently exist, which is our fundamental intension. Taken together, our GEA is able to expand the source gamut of sRGB to the destination gamut of Adobe RGB or BT.2020, while at the same time giving the users a lot of degrees of freedom to control the intensity of the saturation ratio subjectively.