The invention discloses a tooth feature point extraction method, and the method comprises the steps: building a three-dimensional
mesh model of a whole row of natural
dentition where a target tooth is located according to the parameters of the whole row of natural
dentition where the target tooth is located; importing the
point cloud information of the two second posterior molars of the whole row of natural
dentition where the target tooth is located, the target tooth, the left adjacent tooth and the right adjacent tooth into a three-dimensional grid model, and establishing a local coordinate
system; segmenting the target tooth by using the projection, and outputting
point cloud data of the half tooth close to the
occlusal surface; performing twice clustering on the half-tooth
point cloud data, and taking out
Gaussian curvature of a result after the second clustering at a corresponding point of the full-tooth point
cloud data; according to the point with the maximum
Gaussian curvature in each
list, tooth tip feature points of the target tooth are determined, projection and sorting are carried out, the point which is closest to the middle point of the long side and the short side and has the minimum
Gaussian curvature is taken as a tooth socket feature point, all tooth feature points of the target tooth are output, and feature point information of each tooth
package can be accurately extracted through the method.