| CPC H04N 1/6077 (2013.01) [G06N 3/0464 (2023.01); G06N 3/08 (2013.01); G06N 20/00 (2019.01); G06T 3/4046 (2013.01); G06T 7/11 (2017.01); H04N 9/73 (2013.01); H04N 23/88 (2023.01); G06T 2207/10024 (2013.01); G06T 2207/20081 (2013.01)] | 28 Claims |

|
1. A processor-implemented method, the method comprising:
generating, by a neural network model provided an input image, an illumination map comprising illumination values dependent on respective color casts by corresponding plural illuminants, of multiple illuminants affecting the input image, affecting each of a plurality of pixels of the input image; and
generating a white-adjusted image by removing at least a portion of all color casts from the input image using the generated illumination map,
wherein the generating of the illumination map includes generating the illumination map, based on the multiple illuminants, to include a respective illumination value for each pixel in the input image, and
wherein the generating of the white-adjusted image includes:
determining chromaticity information corresponding to each of the multiple illuminants by decomposing the illumination map with respect to each of the multiple illuminants; and
generating, by adjusting the white-adjusted image, a partially white-balanced image with preserved chromaticity information corresponding to a portion of the multiple illuminants, and chromaticity information corresponding to a remaining portion of the multiple illuminants being white information corresponding to the remaining portion of the multiple illuminants.
|