Research on a rapid hidden-line removal and drawing algorithm for large-scale reinforced structures based on geometric parametric representation

With 2D drawings as the vital bridge between designers and constructors, this study proposes a geometric-parametric hidden-line removal algorithm to resolve the long-standing low efficiency and poor accuracy in generating large-scale rebar component drawings across multiple industries and applications. The method represents each bar with a lightweight “central axis + section parameters” model, reducing geometric complexity by transforming 3D solid intersections into parameter-domain analysis and avoiding the high computational cost of B-rep surface intersections. Curvature-driven adaptive triangulation is employed to accurately extract contours of concrete components, while a BVH based coarse-screening and precise-detection pipeline substantially accelerates occlusion computation. To satisfy engineering drawing standards, the algorithm introduces adaptive offset models for orthogonal and oblique intersection scenarios and incorporates refined treatments for bar ends and bends, ensuring consistent double-line width, and smooth geometric transitions. Experiments on 71 components with varying scales demonstrate that the proposed method requires only 10–30 % of the runtime of the OCC algorithm, achieving 67.14–92.16 % efficiency gains, a mean acceleration factor of 18.10, and a 95 % confidence interval of [15.84, 20.44], with stable performance across large-scale assemblies. The generated drawings meet engineering specifications and significantly reduce manual correction. The proposed approach provides an efficient, controllable, and scalable computational framework for automated drawing generation of large-scale rebar components, with strong transferability to bridge reinforcement, rail-transit pipelines, and other slender-structure applications. Future work may explore integrating the parametric centerline–based visibility determination framework—while preserving its core steps and principles—with AI models such as Random Forest, Neural Implicit Fields (NIF) and PolyDiff Model, enabling more efficient and generalizable hidden-line removal and visibility prediction across complex, cross-domain scenarios.