建筑物单体结构化重建的变尺度网格基元提取法

The method of variable scale mesh primitive extraction for monomeric and structural reconstruction of buildings

现有的区域生长算法提取平面基元不准确不规则,忽略了尺度较小的几何结构。本文提出了一种建筑物单体结构化重建的变尺度网格基元提取方法,采用多尺度区域生长算法从网格不同尺度大小的几何结构中提取出对应的平面基元,提升了平面基元提取的准确性和完整性。通过平面基元拓扑优化进一步改善平面基元的拓扑质量,提升多边形表面模型的重建精度;并采用面积优先级策略提高共面平面基元合并效率。试验结果表明,本文方法能够更准确地进行平面基元的提取,保留了建筑物较小尺度的几何结构,生成了更简洁紧凑、结构化的建筑物多边形表面模型。

The existing region-growing algorithm extracts planar primitives inaccurately and irregularly,ignoring the geometric structures with smaller scales.The method of variable scale mesh primitive extraction for monomeric and structural reconstruction of buildings is proposed in this paper,which uses a multi-scale region-growing algorithm to extract the corresponding planar primitives from the geometric structures of different scale sizes of the mesh,improving the accuracy and completeness of planar primitive extraction.The topological quality of planar primitives is further improved by topology optimization of planar primitives,and the reconstruction accuracy of polygonal surface model is enhanced.And the area-priority strategy is used to improve the merging efficiency of co-planar primitives.The experimental results show that this method can extract the planar primitives more accurately,preserve the small-scale geometry of the buildings,and generate a more compact and structural polygonal surface model of the buildings.