ZHAO Qian, REN Zhiqi

(School of Communication and Information Engineering, Xi'an University of Science and Technology, Xi'an 710054, Shaanxi, China)

Abstract: The key point of 3D scanning technology used in tile flatness detection is to extract the coordinates of tile feature points quicklyfrom the tile point cloud. Therefore, a fast title feature point detection algorithm based on tile convex hull of point cloud projection isproposed. First, the tile point cloud is mapped from the three-dimensional space to the two-dimensional plane. Secondly, the convex hull model in computational geometry is introduced to detect the tile corner projection rapidly. Based on this, K-nearest neighbor algorithm is used to obtain the edge feature point projection and center point projection, and finally the two-dimensional plane feature point projection is reflected onto the threedimensional space where the tile feature point coordinates can be obtained. Experiments show that the algorithm can quickly detect the feature points of the tile plane and is of great significance to the automatic detection of the tile flatness by the 3D scanning technology.

Key words: tile flatness; point cloud projection; convex hull