Synthesizing a Reference Image from the Macro Images of a Painting for Vignetting Correction.

Vignetting is an unwanted reduction in the brightness of an image toward the edges compared to the image center caused by camera settings or lens limitations. Conventional vignetting correction methods based on reference images have shown satisfying results. We cannot however use the conventional methods when the reference images are not available. There have been several studies to estimate the vignetting function from microscopic images without any reference images. Most of those studies are based on the fact that it is easy to distinguish foreground and background pixels in a microscopic image. In this paper, we propose a simple method for synthesizing a virtual reference image from macro painting images to carry out vignetting correction. It creates a virtual reference image from many macro images, as obtained by imaging a gray target. Then we apply the existing reference image-based vignetting correction algorithm as it is. According to our experimental results, the proposed method produced satisfying results that are comparable to the results from the conventional methods using a real reference image. Furthermore, the proposed method could more consistently produce the corrected images for unseen input images.