Triangular Mesh Generation for a Real-structure Scan Data in 3D Space

摘要

With the wide application of the areal structured light scanner, the triangulation of the area-structure point cloud become the main bottleneck of the process. Since a one-to-one relationship is maintained between the image pixel and the data point in the area-structure point cloud, the information of the image implicating the certain topology structure can be merged into the point cloud. In order to improve the search speed, a hash table is applied to store the point cloud. A 2D diagonal criterion is derived from the circle criterion, and after introduction of the deformation degree, this criterion can be extended to the 3D space. If the deformation degree is less than a certain threshold, the 3D diagonal criterion can be used, otherwise, the smoothness criterion need to be employed. The proposed algorithm is capable of handling surfaces with complex topology, non-convex boundaries and various interior holes. Experimental results are demonstrated that it is highly efficient compared with other existing algorithms.