Region-based painting style transfer

Abstract

In this paper, we present a novel method for creating a painted image from a photograph using an existing painting as a style source. The core idea is to identify the corresponding objects in the two images in order to select patches more appropriately. We automatically make a region correspondence between the painted source image and the target photograph by computing color and texture feature distances. Next, we conduct a patch-based synthesis that preserves the appropriate source and target features. Unlike previous example-based approaches of painting style transfer, our results successfully reflect the features of the source images even if the input images have various colors and textures. Our method allows us to automatically render a new painted image preserving the features of the source image.

Evaluation of an old-fashioned style picture transformation system using generative adversarial networks and instance segmentation

“…Yamaguchi et al [3] created a system that automatically converted live-action images into an animation style in order to reduce the burden of hand-drawn frames. The system performs style transformation by segmenting regions of the image such as the sky and ground, and then transcribes the corresponding textures.The results showed that the system reflected the style of the animated image better than the conventional method, but there was a point where some textures could not be transferred due to the lack of corresponding parts.…”

Stroke Correspondence by Labeling Closed Areas

“…Constructing closed area correspondences between two raster images has been thoroughly investigated [20]. In the field of reference-based colorization of line drawings, Sato et al [21] utilize the positional and connection relationships of closed areas in the keyframes to construct their correspondence.…”