ABSTRACT

ABSTRACT: In response to the rigid application conditions and limited correction effect intrinsic to classic white balance algorithms, an improved specular reflection algorithm is introduced which divides an entire image into a number of regional blocks, looks for white dots in each of these blocks, derives the estimated values of unknown light sources by calculating the average color information of the dots and eventually corrects the white balance of the image. Our algorithm makes full use of the abundant color information over the image and results in highly reliable light source information. Experiment demonstrates that our algorithm provides an ideal effect of correction.