Feature Extraction Of Facial Point Cloud Data For Intelligent Spectacle Frame Fitting

With the development of social informatization and networking,online spectacle frame fitting,a new retail mode,is in the ascendant.On the other hand,3D laser scanning technology is also becoming more and more mature.Compared with traditional image information,the point cloud generated by scanning has the advantage of richer information.People are in urgent need of a fast and effective automatic parameter acquisition technology for facial 3D scanning point cloud.In this paper,a study is carried out on the extraction method of 3D feature parameters for intelligent spectacle frame fitting,including the calculation of geometric parameters of point cloud mesh model,the simplification method of facial point cloud model,as well as the location of facial feature points and parameters extraction of spectacle frame fitting.The main research contents of this paper include:Firstly,this paper studies the calculation methods of geometric parameters of triangular mesh model,and calculates the principal normal,Gaussian curvature,mean curvature,principal curvature and shape index of facial point cloud mesh model.Secondly,this paper studies the edge collapsing and vertices removing mesh simplification method,an adaptive mesh simplification method based on curvatures is proposed,the vertices of the mesh model were grouped according to the curvature value,and vertices are removed according to the probability according to the grouping situation.So that on the premise of preserving the visual features and geometric curvature features,the facial mesh model is simplified.Next,this paper studies the convolution neural network and face alignment,a method to transform the three-dimensional point cloud information mapping into image information is proposed,and the facial key points are predicted through the neural network to calculate the facial parameters required by the spectacle frame fitting.Finally,this paper develops the "Intelligent Spectacle Frame Fitting Parameter Extraction System",and analyzes the feasibility of the facial parameter extraction method proposed in this paper through an example.The results show that the 3D feature parameter extraction method proposed in this paper has a good effect and meets the requirements of the subsequent parametric design of intelligent spectacle frame.