Palette-based Color Transfer between Images

Chenlei Lv,Dan Zhang

2024-05-14

Abstract:As an important subtopic of image enhancement, color transfer aims to enhance the color scheme of a source image according to a reference one while preserving the semantic context. To implement color transfer, the palette-based color mapping framework was proposed. \textcolor{black}{It is a classical solution that does not depend on complex semantic analysis to generate a new color scheme. However, the framework usually requires manual settings, blackucing its practicality.} The quality of traditional palette generation depends on the degree of color separation. In this paper, we propose a new palette-based color transfer method that can automatically generate a new color scheme. With a redesigned palette-based clustering method, pixels can be classified into different segments according to color distribution with better applicability. {By combining deep learning-based image segmentation and a new color mapping strategy, color transfer can be implemented on foreground and background parts independently while maintaining semantic consistency.} The experimental results indicate that our method exhibits significant advantages over peer methods in terms of natural realism, color consistency, generality, and robustness.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

This paper aims to solve the problems of manual setting in palette - based color transfer methods and poor processing effects on images with scattered color distributions. Specifically:

1. **Reduce manual input**: Existing palette color transfer methods usually require some manual input, which reduces their practicality. This paper proposes a new palette color transfer framework that requires no manual input at all, improving the automation and practicality of the method.

2. **Improve the accuracy and robustness of palette generation**: Traditional methods will experience a significant performance decline when processing images with more dispersed color distributions. In this paper, by redesigning the palette clustering method and using histogram analysis to generate palettes in the Lab color space, the accuracy and robustness of palette generation are improved.

3. **Maintain semantic consistency**: Traditional color transfer methods rely on the semantic correspondence between images. Once this correspondence is lacking, the performance will decline significantly. This paper combines the image segmentation technology of deep learning to independently perform color transfer on the foreground and background parts while maintaining semantic consistency, making the method applicable to images without strict semantic correspondence.

4. **Optimize illumination**: In order to further improve the naturalness of color transfer, this paper introduces an illumination optimization step, including weighted update based on color mapping and global illumination enhancement, to avoid the influence of abnormal exposure.

In summary, the main contribution of this paper is to propose a new palette color transfer framework. This framework can not only automatically generate palettes but also independently perform color transfer on the foreground and background parts while maintaining semantic consistency, and further improve the naturalness of the transfer results through illumination optimization.