一种基于PSFS的近红外人脸3D形貌重构系统  

A near infrared system of 3D facial shape recovery based on PSFS

作　　者：王国珲[1] 吴二星 王立平 WANG Guohui;WU Ering;WANG Liping(School of Optoelectronic Engineering,Xi'an Technological University,Xi'an 110021,China)

机构地区：[1]西安工业大学光电工程学院,陕西西安710021

出　　处：《电视技术》2018年第8期36-41,共6页Video Engineering

基　　金：国家自然科学基金(61102144);西安工业大学陕西省光电测试与仪器技术重点实验室开放基金(2016SZSJ-60-1)

摘　　要：针对非侵犯式人脸3D形貌重构的需求,本文设计了一种基于PSFS的近红外人脸3D形貌重构系统。分析了人脸皮肤在近红外成像条件下的光学属性,系统中利用Oren-Nayar光照明模型来表征其光学属性;相机采集近红外光源照明下的人脸2D图像,结合成像过程中的透视几何变换,构建了2D图像明暗变化与人脸3D形貌相对应的偏微分辐照度方程;该方程可转换为3D形貌梯度函数的二次方程,求解此方程得到H-J类辐照度方程;建立了求解H-J类方程的数值算法,最终获得人脸的3D形貌。实验结果表明,该系统能够有效地实现人脸3D形貌重构,并具有较高的重构精度。Aiming at tlie requirement of tlie non-intrusive 3D facial shape recovery,this paper designs a near infrared system of 3D facial shape recovery based on PSFS.The optical property of tlie facial slcin which is characterized by the Oren-Nayar light il-lumination model is analyzed under near infrared imaging condition.The camera acquires the 2Dimage of the facial surface that is irradiated by a near infrared light source.The partial differential irradiance equation corresponding to the relationship between the 2 Dimage shading variations and the 3 Dfacial shape is constructed with the perspective geometry transformation during the imaging process.Then,the equation can be converted to a quadratic equation of thie function of thie 3D shape gradient and a H-J equation can be obtained by solving the quadratic equation.Finally the 3D facial shape is recovered by establishing a numerical algorithm for approximating the H-J equation.Experimental results illustrate that the recovery systemcan effectively achieve 3D facial shape recovery and has higher recovery accuracy.

关 键 词：PSFS 近红外 人脸3D形貌 Oren-Nayar模型 偏微分辐照度方程 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]