Deep Image Recolorization using Histogram Analogy

Abstract

This thesis presents a novel approach for recoloring a source image with a given reference image. Although there can be diverse pairs of source and reference images in terms of content and composition similarity, previous methods are not capable of covering the whole diversity. To resolve this limitation, we propose a deep neural network that leverages color histogram analogy for image recolorization. A histogram contains essential color information of an image, and our network utilizes the analogy between the source and reference histograms to recolor the source image with the abstract color features of the reference image. In our approach, histogram analogy is exploited basically among the whole images. In the case that the source and reference images have similar contents, the analogy can be exploited among semantic regions to transfer colors between corresponding objects. Experimental results show that our approach effectively recolors the input images in a variety of settings. We also demonstrate a few applications of our approach, such as palette-based recolorization, color enhancement, color editing, and video recolorization.