Research On Three-dimensional Face Automatic Modeling Method Based On Feature Segmentation

The face is an important carrier of information communication and emotional expression for human beings.How to achieve a realistic 3D face reconstruction has always been a research hotspot in the field of computer vision,computer graphics,pattern recognition and image processing.Therefore,the research of 3D face reconstruction with realistic sense has crucial scientific research values and application significance.The core problem of 3D face reconstruction is the face's real sense processing as its quality directly determines the generalization of the 3D face.Based on the deformation model technology,this paper carries out the three-dimensional facial reconstruction,which is divided into two steps,including the establishment of the deformation model and the reconstruction of 3D face.During the deformation model building,the mesh re-sampling method or optical flow method is put forward to align 3D faces,destroying the original geometric features of faces and the degree of automation turns lower.According to this,this paper puts forward the TPS face alignment algorithm of the mesh re-sampling method and then constructs the 3D face deformation model.In 3D face shape reconstruction,two dimensional image and 3D model fitting algorithm has many reconstruction parameters and high cost function,which results in poor realism and low efficiency of 3D face.Thus,this paper proposes the regional fitting algorithm on the basis of the sparse model.Thanks to the above improvements,the authenticity and the degree of automation of 3D face reconstruction are improved.The main innovations of this paper are as follows:First,in this paper,a brand new 3D face reconstruction process is proposed based on the deformation model.With the advantages of thin plate splines,SDM feature detection and geographic division,it constructs a fully automated and high precision face shape reconstruction algorithm.Compared with the traditional method,it has the characteristics of high precision,high efficiency and low cost.Second,research the 3D face alignment technology and put forward the TPS face alignmentalgorithm of the mesh re-sampling method are mainly based on the feature extraction of SDM and the thin-plate spline interpolation function,realizing that the alignment of 3D data is converted to that of the two dimensional data.Third,research the fitting technique of two-dimensional image and 3D model,propose the 3D face regional fitting algorithm based on the sparse deformation model,namely,construct a threedimensional sparse face database to represent the target face and use the least squares to solve the fitting parameters of sparse model.Then,construct the local correspondence between the sparse face and the average face and interpolate by a three-dimensional thin plate spline function to construct a dense 3D face model.Fourth,based on the above research,a 3D face reconstruction system is designed and implemented in this paper.The main functions of the system include the data management and alignment of 3D face,face detection,feature extraction,module division,shape reconstruction and texture mapping.The experiment shows that the system designed in this paper can reconstruct a three-dimensional face model with realistic sense.