Color Constancy Utilizing Standard Deviation Of Color Channels
In computer vision systems color constancy is plays a major role, color constancy technique is generally depends
on spectral distribution of uniform light sources are across the scenes. In this paper we will propose a realistic scenario
where the uniform light sources are too restrictive assumptions. The methodology is proposed to extension of existing
method applying local color constancy to image patches, rather than on entire image globally. Many algorithms have to
estimating the color constancy of the light sources are developed, but all existing methods are applied on single light source
i.e they are consider an image is effected only single uniform illumination, this case is not every time, because images are
effected more than one illuminations. The illumination of single light source is violated by illumination of multiple sources
of light. Here we will discuss a new methodology it considers an image is affected by various light sources, without any
information about the color of light sources. The technique grid based sampling and keypoint based sampling along with
grey edge algorithm can be used to estimate the color of multiple lights. Finally the color correction gives the most
promising and also it has to provide consistency of these algorithms over various types of images taken from different
datasets. This proposed method gives better results than previous methods for color constancy.
Keywords: Computer vision, Color Constancy, Illuminant estimation.