Semantic-Driven 3D Scene Construction Based on Spatial Relationship and Case-Base

摘要

At present, 3D scene construction technology has been widely used in various fields. However, 3D scene construction is increasing on the human and material costs, and the production process is also complicated. First, we use a semantic analysis method to achieve better Chinese automatic word segmentation, which is bidirectional matching Chinese word segmentation based on N-gram model. Next, to solve the problems of low degree of automation and intelligence, we propose a new method of 3D scene construction based on spatial relationship and case-base. The objects and spatial relationships extracted from the scene description texts form a spatial constraint in the form of a triple, which is stored in the spatial relationship template library. Then the 3D model in the case-base is invoked to build the scene. This method takes the spatial constraint as the smallest module of the scene construction, which not only accelerates 3D scene construction, but also improves the rationality of the scene layout. At last, we apply this method to scene generation in a strategy game, in which the effectiveness and efficiency of the new method are proved.