Automatic Generation of 3-D Solid Models From Orthographic Projections

This paper describes an algorithm to generate 3-D solid models of objects from their orthographic projections. This approach is entirely different from the earlier bottom-up approaches. Instead of generating the 3-D vertices, edges and faces sequentially, this approach considers an object to be the boolean combination of a number of elementary objects, which are 2.5D solids. The elementary objects are formed by decomposing a given view into loops and sweeping each loop perpendicular to the direction of projection. This algorithm can handle objects whose projections have curved edge(s) in one view of the orthographic projection. It is believed that proper extension of this algorithm can handle sectional views and multiple views, which has not been addressed in the bottom-up approach.