Research Basis and System Design for in situ Image Diagnosis of Wood in Historic Architecture

The traditional approaches to inspecting timber in ancient structures often entail destructive sampling and chemical assessments, not only undermining the structural integrity, but also consumming considerable time and resources. Consequently, achieving swift, precise, and comprehensive inspections poses a significant challenging. In light of this, in situ image has emerged as a vital information source, enabling the assessment of ancient wood without causing damage. Both domestic and international scholars have extensively researched digital image models for wood diagnosis, paving the way for the diagnosis of various properties of wood applied in historic architecture through in situ image analysis. This paper introduces the design concept of an intelligent diagnosis system for wood in historic architecture. The system is structured into functional modules dedicated to image diagnosis and auxiliary modules, fostering an interactive dynamic between users, platforms, and the backend. The functional modules are sequentially designed, encompassing sample processing, image acquisition, image segmentation, database segmentation and classification, model construction, and deployment. These modules are seamlessly integrated through an image data sharing framework. Adhering to the overall design blueprint, the system will utilize a server as its hardware backbone, integrating specialized software functionalities via API integration. Considering the current research progresses, there are three critical areas requiring further exploration and implementation to expand wood recognition capabilities, minimize the cost and burden of image acquisition, and facilitate system industrialization.