Max-RGB Based Colour Constancy Using the Sub-blocks of the Image

Abstract

Colour constancy refers to the task of revealing the true colour of an object despite ambient presence of intrinsic illuminant. The performance of most of the existing colour constancy algorithms are deteriorated when image contains a big patch of uniform colour. This paper presents a Max-RGB based colour constancy adjustment method using the sub-blocks of the image to significantly reduce the effect of the large uniform colour area of the scene on colour constancy adjustment of the image. The proposed method divides the input image into a number of non-overlapping blocks and computes the Average Absolute Difference (AAD) value of each block's colour component. The blocks with the AADs greater than threshold values are considered having sufficient colour variation to be used for colour constancy adjustment. The Max-RGB algorithm is then applied to the selected blocks' pixels to calculate colour constancy scaling factors for the whole image. Evaluations of the performance of the proposed method on images of three benchmark datasets show that the proposed method outperforms the state of the art techniques in the presence of large uniform colour patches.