基于面域理解的多面体三维重建

Region Recognition Based Polyhedral Model Reconstruction from Orthographic Projections

已有的三维重建算法一般是从点线或体的角度来理解视图 .实际上 ,视图中的环包含了大量的形体面信息 ,可以直接从中间层次的面理解形体的投影 .文中根据实体构成和平面投影的约束和规则重建显式表达的面 ,并进一步推导出隐式表达的面 ,从而得到完整的形体边界模型 .本算法适用于平面多面体 ,通过面域作为中介 ,容易实现与模型引导方法的融合 。

Previous algorithms usually recognize orthographic drawings with the matching of point line or solid projections. But loops in the drawings contain a lot of information about surfaces. In this paper, an efficient algorithm is proposed to rebuild polyhedral model by constructing its faces using constraints and rules recognize from the shape structure and plane's projection. This algorithm can deal with polyhedral model only, however, with the introduction of face region, it can be combined with model oriented approach to handle shapes composed of quadric surfaces.