Active Colorization for Cartoon Line Drawings.

Researchers

Journal

Modalities

Models

Abstract

In the animation industry, the colorization of raw sketch images is a vitally important but very time-consuming task. This paper focuses on providing a novel solution that semi-automatically colorizes a set of images using a single colored reference image. Our method is able to provide coherent colors for regions that have similar semantics than the reference image. An active-learning-based framework is used to match local regions, followed by a Mixed-Integer Quadratic Programming (MIQP) that considers the spatial contexts to further refine the matching results. We efficiently utilize user interactions to achieve high accuracy in the final colorized results. Experiments show that our method outperforms the current state-of-the-art deep learning based colorization method in terms of color coherence with the reference image. The region matching framework could potentially be applied to other applications, such as color transfer.