Multiscale style transfer based on a Laplacian pyramid for traditional Chinese painting

2025-02-07

Electronic Research Archive

AI Summary

Existing style transfer methods—relying on Western oil painting priors and single-scale modeling—suffer from structural distortion and inadequate preservation of aesthetic characteristics when stylizing Chinese traditional paintings (e.g., ink wash and gongbi). To address this, we propose the first two-stage, multi-scale style transfer framework explicitly designed for Chinese painting aesthetics. Our method employs a Laplacian pyramid to enable cross-scale feature disentanglement and reconstruction. A novel Edge Information Selection (EIS) module is introduced to hierarchically model core aesthetic elements—including brushwork intention, negative space, ink gradation, flying-white strokes, and the yin-yang interplay of solidity and void—while preserving structural integrity. Integrated with a base style transfer network and a detail enhancement network, our framework achieves edge-aware multi-scale feature fusion. Extensive experiments on multiple Chinese painting datasets demonstrate significant improvements over AdaIN, WCT, and LapStyle, yielding outputs with superior content fidelity and authentic Chinese artistic style.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Natural Language Processing: Sentiment Analysis, Stylistic Analysis, and Argument Mining
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Style transfer is adopted to synthesize appealing stylized images that preserve the structure of a content image but carry the pattern of a style image. Many recently proposed style transfer methods use only western oil paintings as style images to achieve image stylization. As a result, unnatural messy artistic effects are produced in stylized images when using these methods to directly transfer the patterns of traditional Chinese paintings, which are composed of plain colors and abstract objects. Moreover, most of them work only at the original image scale and thus ignore multiscale image information during training. In this paper, we present a novel effective multiscale style transfer method based on Laplacian pyramid decomposition and reconstruction, which can transfer unique patterns of Chinese paintings by learning different image features at different scales. In the first stage, the holistic patterns are transferred at low resolution by adopting a Style Transfer Base Network. Then, the details of the content and style are gradually enhanced at higher resolutions by a Detail Enhancement Network with an edge information selection (EIS) module in the second stage. The effectiveness of our method is demonstrated through the generation of appealing high-quality stylization results and a comparison with some state-of-the-art style transfer methods. Datasets and codes are available at https://github.com/toby-katakuri/LP_StyleTransferNet.

Problem

• Enhance style transfer for Chinese paintings.
• Address multiscale image information in training.
• Improve detail and structure in stylized images.

Innovation

• Laplacian pyramid decomposition and reconstruction
• Multiscale style transfer method
• Detail Enhancement Network with EIS