Topology optimization is widely used in lightweight structural design to determine optimal material distributions. However, density-based results are represented in an implicit pixel-wise form with blurred boundaries and jagged contours, which limits their direct use in engineering design and manufacturing. This study proposes a two-stage post-processing framework to reconstruct topology optimization results into explicit parametric geometries while preserving structural performance. The framework first extracts and processes contour points from the optimized density field and reconstructs the geometry using Non-Uniform Rational B-Splines (NURBS). A subsequent shape optimization step based on the fixed-grid finite element method (FG-FEM) adjusts boundary control points to reduce performance deviation introduced during reconstruction while satisfying volume and topological homeomorphism constraints. Numerical examples, including the cantilever beam, Michell beam, half-MBB beam, and a quadcopter frame, validate the effectiveness of the framework. The results show that the proposed method enables explicit geometric reconstruction while maintaining structural performance, with compliance deviations within 0.5%–2.6% in benchmark cases.
Tang et al. (Thu,) studied this question.