Illumination Variation in Images in Independent Component Analysis and Principal Component Analysis Subspaces

Abstract

We have investigated the illumination variation in the images taken under a wide variety of lighting conditions in independent component analysis (ICA) and principal component analysis (PCA) subspaces. In ICA subspace, an independent component (IC) is proved to be the composite factor of illumination variation. In PCA subspace, almost all the principal components (PC?s) are proved to contain illumination variation, and a 15-D and 50-D in PCA subspace have more than 5% and 2% related deviation of illumination variation in images, respectively.