【摘要】 针对单幅图像几何结构复杂的组合式建筑建模,为克服单幅图像提供的信息量少、建筑物部件之间相互遮挡关系等局限性,提出了一种基于长方体和圆柱体几何基元定标的组合式建筑交互建模方法。首先对建筑物每个几何部件单独进行场景定标与建模。对长方体部件,用户标注图像投影的6个顶点,系统根据顶点坐标计算相机参数,利用顶点与三维几何体的对应关系构建长方体部件模型;对圆柱体部件,用户标注投影轮廓的4个顶点,计算灭点坐标并利用灭点性质求解相机参数,再构建圆柱体部件模型。然后,利用投影误差和部件几何约束关系引进能量函数对部件进行联合优化生成三维模型,实现组合式建筑的精确重建。最后,利用建筑物纹理的对称信息合成真实的三维组合式建筑模型。实验结果表明:该方法交互简单、建模效率高且扩展性好,可以满足不同复杂结构的组合式建筑物的建模需求。更多还原

【Abstract】 In view of the complex geometric structure of single image, a modular architectures interactive modeling method using geometric primitive calibration is proposed to overcome the limitations of the information provided by a single image and the mutual occlusion between building components. Firstly, the geometric calibration and modeling are carried out separately for each geometric part of the building. For the cuboid component, the user annotates the 6 vertices of the image projection. The system calculates the camera parameters according to the vertex coordinates and constructs the rectangular component model using the corresponding relationship between the vertex and the 3 D geometry. For the cylinder part, the user annotates the 4 vertices of the projection outline. The vanishing point coordinates can be calculated and the camera parameters by using the property of the vanishing point can be solved. So the cylinder component model can be built. Then, using the projection error and the geometric constraint relation of the components, the energy function is introduced into the joint optimization to generate the best three-dimensional scene to achieve the accurate reconstruction of the composite architecture. Finally, the symmetrical information of the building texture is used to synthesize the real three-dimensional combined building model. The experimental results show that the method is simple in interaction, high in modeling efficiency and good in scalability, and can meet the modeling needs of composite buildings with different complex structures.更多还原