重庆理工大学学报（自然科学版）2018，Vol.32Issue(4)：161-170,10.DOI:10.3969/j.issn.1674-8425(z).2018.04.025
基于2维正面人脸图像的3维人脸重构设计与实现
Design and Implementation of 3D Face Reconstruction Based on 2D Frontal Face Image
摘要
Abstract
In order to realize 3 dimensional(3D)face reconstruction based on 2 dimensional(2D) frontal face image,this paper adopts a 3D face reconstruction method of shape and texture mapping based on the 2D frontal face image.Firstly,face image preprocessing,feature point extraction,2D face image pretreatment,and the coordinates and texture information are all obtained by through the active shape model and the explicit shape regression algorithm,that is,there are 3 1 performance feature information of the feature points are marked in the eyes,eyebrows,nose,mouth and other parts of the 2D frontal face image.Then,the standard 3D face model Candide-3 will be used to set up the mapping relationship between the dense 2D face image and the 3D face model by corresponding the 3 1 feature points accurately mapped on the 3D model,so that the 3D model with only statistical characteristics and the standard 3D face model without non personal characteristics becomes rich individuality and authenticity.Then,the face information extracted from the adaptation process,and the 3D model fits the cover with the texture mapping method,and the 3D face model is highly similar to the real human face obtained.We used texture mapping method to cover the 3D model after adaptation,so as to obtain 3D face model with the real face height.Finally,Open GL is used to the 2D image as texture mapping to the target 3D face model,and the 3D face model is deformed into a real 3D face.The experimental results show that the 3D face model interpolation method used in this paper remains similar with the original shape of the generic face model,and which has good smoothness and continuity.关键词
3D人脸重建/主动形状模型/特征提取/径向基函数/纹理映射Key words
3D face reconstruction/active shape model/feature extraction/radial basis function/texture mapping分类
信息技术与安全科学引用本文复制引用
高昊烨,汤艺璇,陈江源,叶超,肖南峰..基于2维正面人脸图像的3维人脸重构设计与实现[J].重庆理工大学学报（自然科学版）,2018,32(4):161-170,10.基金项目
国家自然科学基金资助项目(61573145) （61573145）
广东省公益研究与能力建设专项资金项目(2014B010104001) （2014B010104001）
广东省自然科学基金项目(2015A030308018) （2015A030308018）
