3D Head Reconstruction And Algorithm Research

With the development of Virtual Reality and Augment Reality technology in recent years,the requirement for reconstruction of people based on photos has gradually increased.At present,the practical 3D face reconstruction technology is mainly based on the 3D Morphable Model(3DMM)method.However,due to the cumbersome collection and production of the 3DMM model,the 3DMM model only has a facial part model and cannot reconstruct a complete 3D human head,so it is difficult to apply to the actual engineering.Based on these shortcomings of the 3DMM method,this paper attempts to make a manual 3DMM model.The method of making 3DMM model in this paper is less difficult to make than the traditional 3DMM model.It adopts the common 3D format and can be developed and used by mainstream 3D modeling software.The size of the model is reduced,and it can be easily deployed on samrt iphone.In this thesis,based on the 3DMM model,a complete 3D human head reconstruction algorithm is implemented,including face detection,face feature point extraction,face pose estimation,face shape fitting,face texture extraction,and other algorithms.Experiments show that the above algorithm can reconstruct a realistic three-dimensional human head.The average reconstruction speed on the i7-4790 CPU is only 700 ms,and the reconstruction speed is faster.The traditional face reconstruction can realize the reconstruction of the face by directly using the face part of the photo as a texture.For head reconstruction,this paper uses a pre-made fixed full human head texture mapping.The design algorithm adapts the color of the human head map to the color of the face skin and cal2 culates the optimal affine transformation to transform the face in the photo into the human head map and achieve smooth fusion.Aiming at the actual application scenarios,this paper designs the illumination normalization algorithm to deal with the uneven illumination of the face photos,and improves the reconstruction efficiency of the head under the condition of uneven illumination.In this thesis,the application of 3D head reconstruction is extended,and the 3D head reconstruction software is realized on the mobile phone.By extending the expression model to the 3DMM model,it can fit the expression parameters and realize the facial expression transfer function,and optimize the face pose when the expression is moved,and reduce the face jitter problem when the expression is moved.Under the condition of 3DMM model and algorithm,this paper implements the 50 fps emoticon migration function on the PC and the 30 fps facial expression transfer on the mobile phone.