结合像素模式和特征点模式的实时表情识别

目的 目前2D表情识别方法对于一些混淆性较高的表情识别率不高并且容易受到人脸姿态、光照变化的影响,利用RGBD摄像头Kinect获取人脸3D特征点数据,提出了一种结合像素2D特征和特征点3D特征的实时表情识别方法。方法 首先,利用3种经典的LBP（局部二值模式）、Gabor滤波器、HOG（方向梯度直方图）提取了人脸表情2D像素特征,由于2D像素特征对于人脸表情描述能力的局限性,进一步提取了人脸特征点之间的角度、距离、法向量3种3D表情特征,以对不同表情的变化情况进行更加细致地描述。为了提高算法对混淆性高的表情识别能力并增加鲁棒性,将2D像素特征和3D特征点特征分别训练了3组随机森林模型,通过对6组随机森林分类器的分类结果加权组合,得到最终的表情类别。结果 在3D表情数据集Face3D上验证算法对9种不同表情的识别效果,结果表明结合2D像素特征和3D特征点特征的方法有利于表情的识别,平均识别率达到了84.7%,高出近几年提出的最优方法4.5%,而且相比单独地2D、3D融合特征,平均识别率分别提高了3.0%和5.8%,同时对于混淆性较强的愤怒、悲伤、害怕等表情识别率均高于80%,实时性也达到了10~15帧/s。结论 该方法结合表情图像的2D像素特征和3D特征点特征,提高了算法对于人脸表情变化的描述能力,而且针对混淆性较强的表情分类,对多组随机森林分类器的分类结果加权平均,有效地降低了混淆性表情之间的干扰,提高了算法的鲁棒性。实验结果表明了该方法相比普通的2D特征、3D特征等对于表情的识别不仅具有一定的优越性,同时还能保证算法的实时性。     

基于特征点表情变化的3维人脸识别

目的 为克服表情变化对3维人脸识别的影响,提出一种基于特征点提取局部区域特征的3维人脸识别方法。方法 首先,在深度图上应用2维图像的ASM(active shape model)算法粗略定位出人脸特征点,再根据Shape index特征在人脸点云上精确定位出特征点。其次,提取以鼻中为中心的一系列等测地轮廓线来表征人脸形状;然后,提取具有姿态不变性的Procrustean向量特征(距离和角度)作为识别特征;最后,对各条等测地轮廓线特征的分类结果进行了比较,并对分类结果进行决策级融合。结果 在FRGC V2.0人脸数据库分别进行特征点定位实验和识别实验,平均定位误差小于2.36 mm,Rank-1识别率为98.35%。结论 基于特征点的3维人脸识别方法,通过特征点在人脸近似刚性区域提取特征,有效避免了受表情影响较大的嘴部区域。实验证明该方法具有较高的识别精度,同时对姿态、表情变化具有一定的鲁棒性。     

结合加权局部旋度模式的3维人脸表情识别

目的 3维人脸的表情信息不均匀地分布在五官及脸颊附近,对表情进行充分的描述和合理的权重分配是提升识别效果的重要途径。为提高3维人脸表情识别的准确率,提出了一种基于带权重局部旋度模式的3维人脸表情识别算法。方法 首先,为了提取具有较强表情分辨能力的特征,提出对3维人脸的旋度向量进行编码,获取局部旋度模式作为表情特征;然后,提出将ICNP（interactive closest normal points）算法与最小投影偏差算法结合,前者实现3维人脸子区域的不规则划分,划分得到的11个子区域保留了表情变化下面部五官和肌肉的完整性,后者根据各区域对表情识别的贡献大小为各区域的局部旋度模式特征分配权重;最后,带有权重的局部旋度模式特征被输入到分类器中实现表情识别。结果 基于BU-3DFE 3维人脸表情库对本文提出的局部旋度模式特征进行评估,结果表明其分辨能力较其他表情特征更强;基于BU-3DFE库进行表情识别实验,与其他3维人脸表情识别算法相比,本文算法取得了最高的平均识别率,达到89.67%,同时对易混淆的“悲伤”、“愤怒”和“厌恶”等表情的误判率也较低。结论 局部旋度模式特征对3维人脸的表情有较强的表征能力; ICNP算法与最小投影偏差算法的结合,能够实现区域的有效划分和权重的准确计算,有效提高特征对表情的识别能力。试验结果表明本文算法对3维人脸表情具有较高的识别率,并对易混淆的相似表情仍具有较好的识别效果。     

应用深度光学应变特征图的人脸活体检测

目的 随着人脸识别系统应用的日益广泛,提高身份认证的安全性,提升人脸活体检测的有效性已经成为迫切需要解决的问题。针对活体检测中真实用户的照片存在的人脸欺骗问题,提出一种新的解决照片攻击的人脸活体检测算法。方法 利用局部二值模式LBP（local binary pattern）、TV-L1（total variation regularization and the robust L1 norm）光流法、光学应变和深度网络实现的人脸活体检测方法。对原始数据进行预处理得到LBP特征图;对LBP特征图提取光流信息,提高对噪声适应的鲁棒性;计算光流的导数得到图像的光学应变图,以表征相邻两帧之间的微纹理性质的微小移动量;通过卷积神经网络模型（CNN）将每个应变图编码成特征向量,最终将特征向量传递给长短期记忆LSTM（long short term memory）模型进行分类,实现真假人脸的判别。结果 实验在两个公开的人脸活体检测数据库上进行,并将本文算法与具有代表性的活体检测算法进行对比。在南京航空航天大学（NUAA）人脸活体检测数据库中,算法精度达到99.79%;在Replay-attack数据库中,算法精度达到98.2%,对比实验的结果证明本文算法对照片攻击的识别更加准确。结论 本文提出的针对照片攻击的人脸活体检测算法,融合光学应变图像和深度学习模型的优点,使得人脸活体检测更加准确。     

基于Candide-3算法的图像中面部替换技术

提出了一种基于CANDIDE-3算法的人脸替换算法,该算法使用CLM作为面部特征点定位算法,将2维的面部特征点和CANDIDE中3维的顶点相对应,建立相应的源人脸算法。根据面部特征点估计出头部姿态和表情相关参数,对源人脸进行方向和角度调整,通过颜色转移算法,将源人脸的色彩转换为目标人脸的色彩,然后利用图像融合算法进行融合。实验结果表明,本文的方法能够在图像和视频中进行有效的进行人脸替换。     

网格修补的特征配准方法

目的 网格重建和编辑会产生几何特征缺失的模型,填补这些空洞具有重要的意义。为了克服复杂曲面修补中网格融合难以配准的问题,提出了环驱动球坐标结合基于曲率及法向ICP(iterative closest point)迭代配准的网格修补方法。方法 首先用户查找合适的源网格面片放入空洞处周围;然后对目标网格空洞环建立B样条曲线,将带修补网格包边界置于B样条曲线上,构架环驱动球坐标,将源网格变形初步配准目标网格空洞周围领域;最后使用Laplacian光顺并基于网格曲率及法向进行ICP迭代配准,使源网格与目标网格光滑拼接融合。结果 该方法能够有效修补网格空洞缺失的细节特征,并且拼接处光滑连续。 结论 环驱动球坐标配准避免了网格变形的包围网格笼子构造,再通过ICP迭代精确配准网格,和以往的网格修补方法相比,该方法能够很好地修补网格空洞处细节特征。     

基于协作表示残差融合的3维人脸识别

目的 针对2维人脸难以克服光照、表情、姿态等复杂问题,提出了一种基于协作表示残差融合的新算法.方法 协作表示分类算法是将所有类的训练图像一起协作构成字典,通过正则化最小二乘法代替1范数求解稀疏系数,减小了计算的复杂度,由此系数重构测试人脸,根据重构误差最小原则,对测试人脸正确分类.该方法首先在3维人脸深度图上提取Gabor特征和Geodesic特征,然后在协作表示算法的基础上融合两者的残差信息,作为最终差异性度量,最后根据融合残差最小原则,进行人脸识别.结果 在不同的训练样本、特征维数条件下,在CIS和Texas 2 个人脸数据库上,本文算法的识别率可分别达到94.545%和99.286%.与Gabor-CRC算法相比,本文算法的识别率平均高出了10%左右.结论 在实时成像系统采集的人脸库和Texas 3维人脸库上的实验结果表明,该方法对有无姿态、表情、遮挡等变化问题具有较好的鲁棒性和有效性.     

刚性区域特征点的3维人脸识别

目的 针对3维人脸识别中存在表情变化的问题,提出了一种基于刚性区域特征点的3维人脸识别方法。方法 该方法首先在人脸纹理图像上提取人脸图像的特征点,并删除非刚性区域内的特征点,然后根据采样点的序号,在人脸空间几何信息上得到人脸图像特征点的3维几何信息,并建立以特征点为中心的刚性区域内的子区域,最后以子区域为局部特征进行人脸识别测试,得到不同子区域对人脸识别的贡献,并以此作为依据对人脸识别的结果进行加权统计。结果 在FRGC v2.0的3维人脸数据库上进行实验测试,该方法的识别准确率为98.5%,当错误接受率（FAR）为0.001时的验证率为99.2%,结果表明,该方法对非中性表情下的3维人脸识别具有很好的准确性。结论 该方法可以有效克服表情变化对3维人脸识别的影响,同时对3维数据中存在的空洞和尖锐噪声等因素具有较好的鲁棒性,对提高3维人脸识别性能具有重要意义。     

多任务学习和对抗学习结合的自发与非自发表情识别

目的 如何提取与个体身份无关的面部特征以及建模面部行为的时空模式是自发与非自发表情识别的核心问题,然而现有的自发与非自发表情识别工作尚未同时兼顾两者。针对此,本文提出多任务学习和对抗学习结合的自发与非自发表情识别方法,通过多任务学习和对抗学习捕获面部行为的时空模式以及与学习身份无关的面部特征,实现有效的自发与非自发表情区分。方法 所提方法包括4部分：特征提取器、多任务学习器、身份判别器以及多任务判别器。特征提取器用来获取与自发和非自发表情相关的特征;身份判别器用来监督特征提取器学习到的特征,与身份标签无关;多任务学习器预测表情高峰帧相对于初始帧之间的特征点偏移量以及表情类别,并试图迷惑多任务判别器;多任务判别器辨别输入是真实的还是预测的人脸特征点偏移量与表情类别。通过多任务学习器和多任务判别器之间的对抗学习,捕获面部行为的时空模式。通过特征提取器、多任务学习器和身份判别器的协同学习,学习与面部行为有关而与个体身份无关的面部特征。结果 在MMI（M&M initiative）、NVIE（natural visible and infrared facial expression）和BioVid（biopotential and video）数据集上的实验结果表明本文方法可以学习出与个体身份相关性较低的特征,通过同时预测特征点偏移量和表情类别,有效捕获自发和非自发表情的时空模式,从而获得较好的自发与非自发表情识别效果。结论 实验表明本文所提出的基于对抗学习的网络不仅可以有效学习个体无关但表情相关的面部中特征,而且还可以捕捉面部行为中的空间模式,而这些信息可以很好地改善自发与非自发表情识别。     

表情识别的图像预处理和特征提取方法研究

表情识别是基于视觉信息将脸部的运动或脸部特征的形变进行分类,包括三部分:脸部定位、脸部特征抽取和表情分类.本文首先使用肤色模型进行脸部定位;对提取出来的人脸进行预处理,然后通过Canny鼻子和人脸形状模型相结合的Canny-AAM方法进行特征点定位;最后利用曲线拟合的方法进行特征提取.基于上述算法建立表情识别平台,经过大样本对实时表情识别验证,结果表明对于不同光照下的实时表情识别具有鲁棒性.     

Bilinear Models for 3-D Face and Facial Expression Recognition

In this paper, we explore bilinear models for jointly addressing 3-D face and facial expression recognition. An elastically deformable model algorithm that establishes correspondence among a set of faces is proposed first and then bilinear models that decouple the identity and facial expression factors are constructed. Fitting these models to unknown faces enables us to perform face recognition invariant to facial expressions and facial expression recognition with unknown identity. A quantitative evaluation of the proposed technique is conducted on the publicly available BU-3DFE face database in comparison with our previous work on face recognition and other state-of-the-art algorithms for facial expression recognition. Experimental results demonstrate an overall 90.5% facial expression recognition rate and an 86% rank-1 face recognition rate.      

基于多尺度分析的自动人脸照片移植

针对人脸移植中输入图像与目标图像的脸部姿态、光照环境与颜色分布不一致的问题,提出了一种基于多尺度分析的自动人脸照片移植方法。通过多线性模型从单张图像中恢复三维人脸模型,从而自动变换输入图像中的人脸姿态。提出了一种多尺度增强与融合算法,根据目标图像的细节特征对输入图像自动调整,并通过无缝融合合成新的人脸照片。实验结果表明该方法可以使输入图像有效匹配目标图像的明暗变化与颜色分布,并自适应调整局部细节。该方法对各种人脸图像之间的移植鲁棒性高,合成照片真实感强。     

Performance‐driven muscle‐based facial animation

We describe a system to synthesize facial expressions by editing captured performances. For this purpose, we use the actuation of expression muscles to control facial expressions. We note that there have been numerous algorithms already developed for editing gross body motion. While the joint angle has direct effect on the configuration of the gross body, the muscle actuation has to go through a complicated mechanism to produce facial expressions. Therefore,we devote a significant part of this paper to establishing the relationship between muscle actuation and facial surface deformation. We model the skin surface using the finite element method to simulate the deformation caused by expression muscles. Then, we implement the inverse relationship, muscle actuation parameter estimation, to find the muscle actuation values from the trajectories of the markers on the performer's face. Once the forward and inverse relationships are established, retargeting or editing a performance becomes an easy job. We apply the original performance data to different facial models with equivalent muscle structures, to produce similar expressions. We also produce novel expressions by deforming the original data curves of muscle actuation to satisfy the key‐frame constraints imposed by animators.Copyright © 2001 John Wiley & Sons, Ltd.     

Real‐Time Facial Expression Transformation for Monocular RGB Video

This paper describes a novel real‐time end‐to‐end system for facial expression transformation, without the need of any driving source. Its core idea is to directly generate desired and photo‐realistic facial expressions on top of input monocular RGB video. Specifically, an unpaired learning framework is developed to learn the mapping between any two facial expressions in the facial blendshape space. Then, it automatically transforms the source expression in an input video clip to a specified target expression through the combination of automated 3D face construction, the learned bi‐directional expression mapping and automated lip correction. It can be applied to new users without additional training. Its effectiveness is demonstrated through many experiments on faces from live and online video, with different identities, ages, speeches and expressions.     

Manifold‐Based 3D Face Caricature Generation with Individualized Facial Feature Extraction

Caricature is an interesting art to express exaggerated views of different persons and things through drawing. The face caricature is popular and widely used for different applications. To do this, we have to properly extract unique/specialized features of a person's face. A person's facial feature not only depends on his/her natural appearance, but also the associated expression style. Therefore, we would like to extract the neutural facial features and personal expression style for different applicaions. In this paper, we represent the 3D neutral face models in BU–3DFE database by sparse signal decomposition in the training phase. With this decomposition, the sparse training data can be used for robust linear subspace modeling of public faces. For an input 3D face model, we fit the model and decompose the 3D model geometry into a neutral face and the expression deformation separately. The neutral geomertry can be further decomposed into public face and individualized facial feature. We exaggerate the facial features and the expressions by estimating the probability on the corresponding manifold. The public face, the exaggerated facial features and the exaggerated expression are combined to synthesize a 3D caricature for a 3D face model. The proposed algorithm is automatic and can effectively extract the individualized facial features from an input 3D face model to create 3D face caricature.     

Multiple Facial Image Editing Using Edge–Aware PDE Learning

This paper introduces a novel facial editing tool, called edge‐aware mask, to achieve multiple photo‐realistic rendering effects in a unified framework. The edge‐aware masks facilitate three basic operations for adaptive facial editing, including region selection, edit setting and region blending. Inspired by the state‐of‐the‐art edit propagation and partial differential equation (PDE) learning method, we propose an adaptive PDE model with facial priors for masks generation through edge‐aware diffusion. The edge‐aware masks can automatically fit the complex region boundary with great accuracy and produce smooth transition between different regions, which significantly improves the visual consistence of face editing and reduce the human intervention. Then, a unified and flexible facial editing framework is constructed, which consists of layer decomposition, edge‐aware masks generation, and layer/mask composition. The combinations of multiple facial layers and edge‐aware masks can achieve various facial effects simultaneously, including face enhancement, relighting, makeup and face blending etc. Qualitative and quantitative evaluations were performed using different datasets for different facial editing tasks. Experiments demonstrate the effectiveness and flexibility of our methods, and the comparisons with the previous methods indicate that improved results are obtained using the combination of multiple edge‐aware masks.     

A deformation model to reduce the effect of expressions in 3D face recognition

In 3D face recognition, most work utilizes the rigid parts of face surfaces for matching to exclude the distortion caused by expressions. However, across a broad range of expressions, the rigid parts may not always be uniform and cover large parts of faces. On the other hand, the non-rigid regions of face surfaces also contain useful information for recognition. In this paper, we include the non-rigid regions besides the rigid parts for 3D face recognition. A deformation model is proposed to deform the non-rigid regions to the shapes that are more similar between intra-personal samples but less similar between inter-personal samples. Together with the rigid regions, the deformed parts make samples more discriminable so that the effect of expressions is reduced. The first part of our model uses the target gradient fields from enrolled samples to depress the distortion of the non-rigid regions. The gradient field works in the differential domain. According to the Poisson equation, a smooth deformed shape can be computed by a linear system. The second part of the model is the definition of a surface property that determines the deformation ability of different face regions. Unlike the target gradient fields that improve the similarity of intra-personal samples, the original topology and surface property can keep inter-personal samples sufficiently dissimilar. Our deformation model can be used to improve existing 3D face recognition methods. Experiments are carried out on FRGC and BU-3DFE databases. There are about 8–10% improvements obtained after applying this deformation model to the baseline ICP method. Compared with other deformation models, the experimental results show that our model has advantages on both recognition performance and computational efficiency.     

双重对偶生成对抗网络的跨年龄素描—照片转换

目的 跨年龄素描-照片转换旨在根据面部素描图像合成同一人物不同年龄阶段的面部照片图像。该任务在公共安全和数字娱乐等领域具有广泛的应用价值,然而由于配对样本难以收集和人脸老化机制复杂等原因,目前研究较少。针对此情况,提出一种基于双重对偶生成对抗网络（double dual generative adversarial networks,D-DualGANs）的跨年龄素描-照片转换方法。方法 该网络通过设置4个生成器和4个判别器,以对抗训练的方式,分别学习素描到照片、源年龄组到目标年龄组的正向及反向映射。使素描图像与照片图像的生成过程相结合,老化图像与退龄图像的生成过程相结合,分别实现图像风格属性和年龄属性上的对偶。并增加重构身份损失和完全重构损失以约束图像生成。最终使输入的来自不同年龄组的素描图像和照片图像,分别转换成对方年龄组下的照片和素描。结果 为香港中文大学面部素描数据集（Chinese University of Hong Kong（CUHK）face sketch database,CUFS）和香港中文大学面部素描人脸识别技术数据集（CUHK face sketch face recognition technology database,CUFSF）的图像制作对应的年龄标签,并依据标签将图像分成3个年龄组,共训练6个D-DualGANs模型以实现3个年龄组图像之间的两两转换。同非端到端的方法相比,本文方法生成图像的变形和噪声更小,且年龄平均绝对误差（mean absolute error,MAE）更低,与原图像相似度的投票对比表明1130素描与3150照片的转换效果最好。结论 双重对偶生成对抗网络可以同时转换输入图像的年龄和风格属性,且生成的图像有效保留了原图像的身份特征,有效解决了图像跨风格且跨年龄的转换问题。     

A Generic Framework for Efficient 2-D and 3-D Facial Expression Analogy

Facial expression analogy provides computer animation professionals with a tool to map expressions of an arbitrary source face onto an arbitrary target face. In the recent past, several algorithms have been presented in the literature that aim at putting the expression analogy paradigm into practice. Some of these methods exclusively handle expression mapping between 3-D face models, while others enable the transfer of expressions between images of faces only. None of them, however, represents a more general framework that can be applied to either of these two face representations. In this paper, we describe a novel generic method for analogy-based facial animation that employs the same efficient framework to transfer facial expressions between arbitrary 3-D face models, as well as between images of performer's faces. We propose a novel geometry encoding for triangle meshes, vertex-tent-coordinates, that enables us to formulate expression transfer in the 2-D and the 3-D case as a solution to a simple system of linear equations. Our experiments show that our method outperforms many previous analogy-based animation approaches in terms of achieved animation quality, computation time and generality.