Rapid Reconstruction of a Three-Dimensional Mesh Model Based on Oblique Images in the Internet of Things

One of the main targets of the Internet of Things (IoT) is the construction of smart cities, and many industries based on the IoT serve popular applications in a smart city. However, 3-D reconstruction constitutes a major difficulty in the construction of a smart city. In recent years, oblique photography technology has been widely applied in the rapid 3-D modeling and other aspects of smart cities. However, in the automatic construction of a 3-D mesh model for oblique photogrammetry, complex building geometries make it very difficult to construct a triangular mesh model. Therefore, a network construction method is needed that can not only effectively construct a 3-D mesh model but also address the results of auto-modeling for an oblique image. The representative network construction method is a huge triangulation network in which the constructed surface of the object does not satisfy the manifold features and it is inconvenient to optimize and edit the model, yielding a low network construction efficiency. To solve these problems, a new method for constructing a high-quality manifold mesh model is proposed in this paper. First, an adaptive octree division algorithm is used to divide the point cloud data into sub domains that cover each other. Then, a mesh reconstruction is performed in each sub domain, and an efficient mesh construction algorithm based on relabeling the vertices of the directed graph is proposed to construct the manifold mesh. Finally, a triangular facet orientation method is used to homogenize the normal vectors of the mesh. The experimental results proof that the proposed method greatly improves the mesh reconstruction, effectively reflects the model details, and possesses a strong anti-noise ability. Also, it has a good robustness and is particularly suitable for the 3-D reconstruction of large scenes and complex surfaces.