Color Space Identification for Image Display

Abstract

Available color images can be encoded in any color space. However, according to the image display model, it is assumed that the color image is encoded in a specific color space belonging to the RGB family. Displaying an encoded image in a color space by using a system designed for the display of encoded images in another color space leads to a poor reproduction of colors. To overcome this problem, the encoded image in a color space must be converted to the color space used by the display system. Unfortunately, the image color space can be not included in the image metadata and therefore it is unknown. Even if the display systems are massively used, this issue does not seem to be tackled before. In this paper, we propose the identification of the image color space from its colors. First, the Gamut of color spaces is estimated by using a collection of images. Then, the image color histogram is compared to the estimated Gamuts. The obtained identification scores using four color spaces and a collection of 2106 images are encouraging.