Multimodal intelligent reconstruction-driven digital regeneration of cultural heritage in Qing Dynasty tea plantation landscapes

Advances in AI, particularly deep learning, are transforming the protection and utilization of historical and cultural landscapes. This study proposes a comprehensive digital regeneration workflow for fragmented historical images that systematically integrates image analysis, including segmentation and color extraction, with conditional generation and dynamic presentation. Based on 104 Qing Dynasty export paintings, we develop a hybrid “Stable Diffusion + ControlNet” architecture that incorporates semantic segmentation and HSV color analysis to achieve fine-grained scene generation. The research establishes a complete multimodal technical process of “intelligent reconstruction–dynamic regeneration”, extending beyond the mere application of generative models. The final output transforms fragmented historical imagery into narrative-driven multimedia spatial demonstrations, offering a reusable operational framework and methodology for immersive heritage experiences and sustainable cultural preservation.