Many recent approaches to the problem of interpreting line drawings as solid objects treat inflation as a two-stage approach, the first stage being to produce a quick initial estimate of vertex zcoordinates, and the second being to refine these initial estimates to produce a “more beautiful” geometry. By making assumptions about engineering objects and the ways people see and depict them, it is often possible to reproduce a single object which humans will agree is the correct interpretation of the drawing. Of these assumptions, those to do with perpendicularity are most important, in part because perpendicularity is the most common regularity in engineering objects, and in part because of the importance of perpendicularity in the human perception process. In this paper, we catalogue various possible instances of perpendicularity in engineering drawings.
Download Full PDF Version (Non-Commercial Use)