There are a lot of layout problems in modern engineering and real life, and indoor spatial layout is one of the very important branches. The indoor spatial layout focuses on the indoor scenes that people rely on for production and life. At the same time, indoor scene models are gradually being applied to more and more fields, such as CG animated movies, computer games, virtual reality technology and so on. Because of its wide application prospect and huge commercial value, indoor space layout has become an important research topic in the field of computer graphics and computer vision. This article focuses on the automatic design and optimization of indoor space layout, and proposes a single sample driven indoor space layout optimization technology based on a given layout boundary or layout space. Given indoor scenes and smaller indoor spaces, this paper proposes a scene redirection method that can reduce the size of indoor scenes to the size that the indoor space can accommodate. The scene redirection method in this article maintains the layout and structure of the indoor scene, reducing the scene size while maintaining spatial perception in the scene. Finally, this article verifies the effectiveness of this method using convolutional neural network (CNN), and the experimental results show that the scene redirection method proposed in this paper is effective.
|