Color Correction for Multi-view Images Combined with PCA and ICA

Color information is very important in setting the style of images, but always influenced by internal and external imaging factors. For this reason, it is necessary to eliminate color inconsistency between different views in multi-view imaging. Up to now, many color transfer methods have been proposed mainly for vivid color appearance, in which the mean, variance, or covariance information are used to transfer color. In this paper, by taking the advantage of the principal component analysis (PCA) and independent component analysis (ICA), a general color correction framework is proposed. We first analyze the correction property for target image and source image. Then we perform global PCA color correction if global correction property satisfied. Otherwise, we separate the independent signal sources by dominant color extraction, and establish color correction relations between independent sources. Finally, we combine the ICA correction results to get the final corrected image. The Experimental results show that well subjective and objective performances can be achieved with the proposed method. Key-Words: Multi-view image, color correction, principal component analysis, independent component analysis, dominant color extraction, CIEDE2000.