Illumination Estimation Based on Color Invariant
-
Abstract
Illumination estimation is most important step in colorconstancy computation. Although a number of elaborate algorithms wereproposed, methods like grey world and Max are still widely usedbecause of their low computational costs. The grey world algorithm isbased on the grey world assumption: the average reflectance in a sceneis achromatic. But this assumption can not be always satisfied well.Borrowing on some of the strengths and simplicity of the grey worldalgorithm, we propose a novel algorithm called Illumination estimationbased on color invariant (IECI) to improve the performance of greyworld, in which the achromatic surfaces are identified for illuminationestimation. As we know, the color of achromatic surface can completelyreflect the illumination color incident on the scene. In light of thisidea, color invariant technique is introduced to approximately identifythe achromatic surfaces, then iterative method is used to improveaccuracy of the proposed method. The experimental results show that theproposed algorithm can get satisfying results and outperforms somecurrent state-of-the-art color constancy algorithms.
-
-