This paper presents an approach for image cartooniza- tion. By observing the cartoon painting behavior and consulting artists, we propose to separately identify three white-box representations from images: the surface rep- resentation that contains a smooth surface of cartoon im- ages, the structure representation that refers to the sparse color-blocks and flatten global content in the celluloid style workflow, and the texture representation that reflects high- frequency texture, contours, and details in cartoon im- ages. A Generative Adversarial Network (GAN) framework is used to learn the extracted representations and to car- toonize images.
The learning objectives of our method are separately based on each extracted representations, making our frame- work controllable and adjustable. This enables our ap- proach to meet artists’ requirements in different styles and diverse use cases. Qualitative comparisons and quanti- tative analyses, as well as user studies, have been con- ducted to validate the effectiveness of this approach, and our method outperforms previous methods in all compar- isons. Finally, the ablation study demonstrates the influence of each component in our framework.