Identifying essential segments and points in 2D image for extracting a realizable 3D shape using rule-based approach

In this paper, we consider a new aspect of the 3D readability problem as follows. Given of a set of lines laying on a 2-dimensional space, is there any 3D shape realizable from these lines. Our studied problem is different from the classical readability problem in the following senses. Firstly, all given lines have no limited length. Both ends can be extended to infinity. They can cross one another to form segments. Secondly, each crossing point can be viewed as a vertex and each line segment can be viewed as an edge of a graph. Some vertices and edges are redundant and useless in the realization. The problem is which vertices and edges are significant and essential in the realization. Our proposed solution was tested with 200 polyhedral object images obtained from P.A.C. Varley´s thesis with 71% accuracy.