An international conference connecting people
in CAD research, education and business
Bookmark and Share
Copyright (C) CAD Solutions, LLC. All rights reserved.
Proceedings of CAD'17, 2017, 420-424
Learnable CAD for Reconstructing 3D Models from Sketches

Masaji Tanaka, Ryosuke Yamaoka, Okayama University of Science

Abstract. Generally sketches as line drawings are important tools for designers when they create new mechanical parts and so forth. A human can draw a sketch of a 3D object, and the other human would recognize the object from the sketch. This human behavior fascinated a lot of researchers who long for realizing artificial intelligence systems. If a system that can automatically reconstruct a 3D object as a solid model from a sketch drawn in CAD is realized, it would be useful for designers and so forth. Since decades, this reconstruction system has been researched by many researchers. Although their proposed methods could automatically reconstruct solid models from sketches, it was difficult for them to handle sketches including curved lines. Recently we proposed a method that could handle curved lines in sketches drawn in CAD, firstly several sketches of simple objects such as a cuboid and a hole were defined as sketch features, and when a sketch of a complex object was input, each sketch feature was detected and extracted repeatedly from the sketch until all lines were deleted in the sketch. After a feature was extracted from a sketch, the sketch would be broken and meaningless. So some restoration process would be required. Also, since the restoration process could exist unlimitedly for many kinds of sketches, we applied our inductive learning technique.    

Keywords. Inductive Learning, Sketch, Line Drawing, Reconstruction, Learning Interface

DOI: 10.14733/cadconfP.2017.420-424