P‐49: A Gamut‐Mapping Algorithm with Separate Skin and Non‐Skin Color Preference Controls for Wide‐Color‐Gamut TV

Abstract

Recently emerging wide‐color‐gamut (WCG) TVs face two challenges. Viewers welcome the increase in color saturation of WCG TVs, but do not welcome unnatural skin color rendering due to the stretching of the narrower color gamut (such as the color gamut defined in sRGB/ITU‐R BT.709/601). In this paper, we propose a color gamut mapping algorithm (GMA□ to address the problem. The proposed GMA consists of two sub‐GMAs: the first sub‐GMA is focused on correctly rendering skin colors, and the second sub‐GMA is on rendering no‐ skin colors. In order to allow viewers to independently adjust skin and non‐skin colors according to their preference, the skin colors rendered by the first sub‐GMA and the non‐skin colors rendered by the second sub‐GMA go through two seperate viewer preference adjustment modules. Then the mapped and viewer‐adjusted skin and non‐skin colors are blended according to skin scores, which are scalars between 0 and 1. A skin score is assigned to a color to describe its likelihood to be a skin color and is obtained by extensive training on sample sets. Skin scores of an image are filtered by a linear lowpass filter to prevent generating artifacts.

Vision Models for Wide Color Gamut Imaging in Cinema

“…A few methods [44], [45] perform gamut extension using functions learned from user studies. Unlike the aforementioned GEAs, some global methods [46], [47], [48] first classify the colors of the input image according to a criterion, and then perform gamut extension differently for each class. For example, labelling each color of a given image as skin or non-skin [46]; dealing with objects of low chroma and high chroma differently [47]; identifying certain memory colors such as green grass and blue sky, and rendering them independently [48].…”

“…Unlike the aforementioned GEAs, some global methods [46], [47], [48] first classify the colors of the input image according to a criterion, and then perform gamut extension differently for each class. For example, labelling each color of a given image as skin or non-skin [46]; dealing with objects of low chroma and high chroma differently [47]; identifying certain memory colors such as green grass and blue sky, and rendering them independently [48]. Other approaches [49], [50] propose three types of extensions: chroma extension, extension along lines from the origin, and adaptive mapping that is a compromise between the first two strategies.…”

Spatial gamut mapping among non-inclusive gamuts

“…Other methods label the data in different categories in order to perform gamut extension differently in each category. For example, [19,20] are defined to preserve skin tones, while [21] performs a different extension depending on the chroma of the object, and [22] categorizes different memory colors. In [6] different methods are proposed and evaluated.…”

Gamut Extension for Cinema

“…Later, Kang et al [21] and Anderson et al [5] presented user assisted methods to deal with the problem of gamut extension. While all these aforementioned algorithms treat each color without analyzing the content of the input image, Pan and Daly [33], Casella et al [12], Heckaman and Sullivan [18] introduced methods that first classify the colors of the input image according to some criterion and then extend them. In particular, the work of [33] labels each color of a given image as skin or nonskin, [12] deals with objects of low chroma and high chroma differently, and [18] identifies certain memory colors such as green grass and blue sky, and renders them independently.…”