Systematic reconstruction of 3D curvilinear objects from two-view drawings

This paper presents an efficient algorithm for reconstructing curvilinear 3D objects from two-view drawings. The main features of the algorithm are to improve the speed of the reconstruction process and to generate a sequence of solutions. First, the partially constructed objects are reconstructed from the restricted candidate faces corresponding to each area in the two-view drawings. Then the complete objects are obtained from the partially constructed objects by adding perpendicular faces with geometric validity. By limiting the number of candidate faces corresponding to areas only, the computational time of the combinatorial search process is considerably reduced. Second, to find more plausible solutions prior to unusual ones, the algorithm generates the sequence number to multiple solutions based on a plausibility function that maps the proposed solution to the designer's intention in the original drawings. Furthermore, the proposed method of reconstruction supports a wide scope of general (manifold and non-manifold) 3D objects containing cylindrical, conical, spherical, and toroidal faces. Several examples from a working implementation are given to demonstrate the completeness of the algorithm.