A Learning Method for Automatic Reconstructing 3D Models from Sketches

Masaji Tanaka, Chiharu Higashino, Tetsuya Asano, Keigo Takasugi

Machine Learning, Reconstruction, Sketch, 3D Model, Line Drawing
Presently solid modelers have become popular tools in CAD/CAM systems. However, much study and training for many operations would be required for a user when he/she makes a solid model of an object smoothly. Although many simpler solid modelers and 3D sketching systems have been developed in recent years, minimum operations would exist and they would become unique. Therefore, they would become very far from standard solid modelers such as CATIA. As the result, a lot of them would be disappeared naturally in the future. On the other hand, the methods to automatically reconstruct solid models from sketches have been researched. Also, we have been developing a system for the reconstruction. Generally sketches are convenient for expressing the shapes of 3D objects with papers. Most people can draw them without knowing drawing rules, and also they are often used when designers discuss new parts, products, etc. If our system realizes, it would become a powerful tool in CAD/CAM systems. In the system, stronger machine learning sub-systems are required and it has been an important issue. In this paper, a new learning method for the issue is proposed. In the system, firstly many simple sketch features are defined. 
CAD’18, Paris, France

Proceedings of CAD’18

2018 CAD Solutions, LLC, http://www.cadconferences.com

422

426