Automatic white balance based on dynamic mapping

摘要

A method of white balancing an image includes mapping pixels of the image to a color space diagram. Each of the pixels of the image include a red (“R”), a green (“G”), and a blue (“B”) subvalue. A first central tendency of each of the RGB subvalues of pixels mapped in a first pre-defined region of the color space diagram is determined and a second central tendency of each of the RGB subvalues of pixels mapped in a second pre-defined region of the color space diagram is determined. The first pre-defined region is associated with a first illuminating source and the second pre-defined region is associated with a second illuminating source. RGB values of a white pixel are generated based on the first and second central tendencies.

主权项

1. A method of white balancing an image comprising:
mapping pixels of the image to a color space diagram, wherein each of the pixels include a red (“R”), a green (“G”), and a blue (“B”) subvalue; determining first central tendencies of each of the RGB subvalues of the pixels mapped in a first pre-defined region of the color space diagram and determining second central tendencies of each of the RGB subvalues of the pixels mapped in a second pre-defined region of the color space diagram, wherein the first pre-defined region is associated with a first illuminating source and the second pre-defined region is associated with a second illuminating source; applying a first weighting factor to each of the first central tendencies and applying a second weighting factor to each of the second central tendencies; generating mean RGB values of a white pixel based on the first central tendencies with the first weighting factor applied and the second central tendencies with the second weighting factor applied; mapping the white pixel to the color space diagram; correcting the white pixel by a correction vector on the color space diagram; and white balancing the image by applying the correction vector to the red and blue subvalues of the image.