A progressive geometric enhancement method for automated recognition of steel bar drawings in prefabricated construction

Prefabricated construction has gained significant attention in the construction industry due to its higher production efficiency and reduced labor demands. However, the transfer of graphical information from steel bar drawings to rebar equipment still largely depends on manual recognition methods. This paper proposes the progressive geometric enhancement method (PGEM) to automate the recognition of rebar drawings using enhanced kernel density estimation and Hough transform. It transforms the problem of graphics recognition into the problem of density estimation and fitting of dot units for automated solving. The automated recognition results from over 480 steel bar drawings demonstrated a accuracy of 97.7% and an average efficiency over 11.8 times higher than manual method. Moreover, a graphical user interface was developed and applied to over 100,000 drawings in collaboration with various construction companies. The application demonstrates PGEM’s potential to replace manual recognition, thereby improving the efficiency and reducing the labor-intensive nature of the drawing recognition process in prefabricated construction.