【摘要】 针对大范围城市场景中建筑物倾斜摄影三维模型存在的噪声明显、规则性缺失、结构粗糙等问题，提出了一套基于轮廓约束的精细化修复方法。首先，利用OSG技术实现建筑物模型的切割单体化，通过空间几何与投影面形状特征提取三角面片顶点并完成纹理映射。其次，基于建筑物墙面相交为棱线的几何特性，对墙面三角面片进行分类，通过最小二乘拟合得到墙面拟合面，提取墙角线并构建完整轮廓。最后，提出改进的拉普拉斯网格平滑算法修复墙面粗糙问题，通过两两相交直接修复棱线，并以最小二乘拟合面为投影特征面，设计基于拟合面的孔洞修复算法修复墙面孔洞。实验结果表明，该方法在建筑物模型的单体化、轮廓构建和精细化修复方面具有较高的可行性和有效性，能够显著提升模型精度和规则性，为智慧城市建设提供了有力的技术支持。更多还原

【Abstract】 Aiming at the problems of obvious noise,lack of regularity,and rough structure in the 3D models of buildings derived from large-scale urban tilted photography,a refined restoration methods based on contour constraints is proposed.Firstly,the OSG technology is used to achieve the cutting and individualization of building models,extracting the vertices of triangular facets through spatial geometry and projection surface shape features,and completing texture mapping. Secondly,based on the geometric characteristics of intersecting building walls as edges,the wall triangles are classified,the wall fitting surface is obtained through least squares fitting,and the corner lines are extracted and the complete contour is constructed.Finally,an improved Laplacian grid smoothing algorithm is proposed to repair rough wall surfaces. The edges are directly repaired by intersecting each other,and a hole repair algorithm based on the least squares fitting surface is designed to repair wall holes using the fitting surface as the projection feature surface. The experimental results show that this method has high feasibility and effectiveness in the individualization,contour construction,and fine repair of building models,and can significantly improve model accuracy and regularity,providing strong technical support for smart city construction.更多还原