A Study of Generating Method of 3D Consistent Picture Based on Defining Object Properties and Spatial Relations

概要

論文の詳細を見る
Conventional CAD systems and 3D drawing tools can easily generate a 3D picture on the computer graphics screen. However, the picture is often misinterpreted by misplacement of objects because these drawing tools do not have the ability to reason about objects and constraints on their position. Therefore, we propose a system that can generate a 3D consistent picture. By defining object properties, the system can reason about the objects, and by maintaining the spatial relations, the object can be placed consistently with respect to the other objects. Thus, a consistent picture is produced as the output. In this paper, the mechanisms of defining object properties are explained in detail. Moreover, the methods used in the learning system to automatically generate a consistent picture are also discussed.
東海大学の論文