An extraction method for structural features based on edge detection and multi-conditional filtering: a case study of the steel box girder from engineering blueprints

A complete set of engineering blueprints often contains a wealth of design experience and a profound comprehension of the structure from the designers. Consequently, the analysis and study of the visual data contained in these blueprints are vital. Structure feature extraction refers to extracting the external contour lines and internal structure layout of the target structure from engineering blueprints. This study proposes the Edge Detection-based and multi-conditional filtering method (EDMFM), an innovative method utilizing Computer Vision Technology. The EDMFM uses image processing techniques to optimize traditional manual extraction methods, which rely heavily on keying software to remove the background of the image, with the disadvantage of time-consuming, labor-intensive and imprecise extraction results. Additionally, this study conducts a qualitative and quantitative analysis of the extraction results from three aspects, followed by ablation experiments on the functional modules of the proposed method. The experimental results demonstrate that the proposed method exhibits good applicability in extracting key structural features from blueprint images, providing a foundation for further exploration and analysis of structural features at an advanced level.