Style and Structure-Aware Generative Super-Resolution for Bashu Painting and Calligraphy Restoration

As a vital component of cultural heritage, traditional Chinese painting and calligraphy require advanced image reconstruction techniques for digital preservation. A major challenge lies in maintaining fine brushwork and complex artistic styles during super-resolution reconstruction, as existing methods often focus on low-level details while neglecting structural and semantic coherence. To address this limitation, this study proposes a generative super-resolution framework based on artistic feature decomposition, incorporating a style embedding module and a structure-guidance mechanism. The method employs a two-stage process: artistic style features are first extracted, followed by semantic-driven reconstruction, enabling a better capturing of the coupling between visual data and artistic expression. Experimental results show that the proposed approach achieves superior performance in clarity, style consistency, and structural accuracy, offering strong potential for application in cultural heritage preservation and digital restoration.