人脸皮肤软组织跨尺度变形测量与表情形成机理研究

The human face, as the most distinctive character of individual identity, usually has six typical kinds of facial expression, i.e., surprise, disgust, anger, fear, sadness, joy, etc. Large-scale facial expression can be qualitatively analyzed by visual observation. Nevertheless, quantitative analysis of skin deformation, as well as micro-expression are becoming the very important research topics. Difficulty of the problem lies in developing in-situ measurement technique for mechanical properties of facial skin. Moreover, a sequential deformation analysis with non-contact and high resolution is imperative. The mechanical properties of skin are important for a number of applications including surgery, dermatology, impact biomechanics and forensic science. The characterization of skin mechanics has many clinical implications and facial animation by using digital images based on the mechanism theory and muscle model, and has been an active area of research for the past few decades. Biomechanical models have evolved from earlier empirical models to state-of-the-art structural models that provide linkage between tissue microstructure and macroscopic stress-strain response. In this project, we will measure and investigate real-time changes in facial expression with the assistance of psychologists. An in-situ measurement device will be developed to study the mechanical properties of facial skin. The constitutive model and transmission mechanism for stress-strain will be studied by means of experiment and finite element method. Additionally, a new displacement function and algorithm will be exploited. Meanwhile, by using multiple-cameras adaptive 3D-DIC parallel algorithm, we will realize regional division and real-time tracking of facial deformation and accurate measurement for scale-span three-dimensional deformation. Furthermore, automatic identification of facial expression feature will come true by processing the deformation data. And then, we can obtain the structure and feature information (including shape and size of structures of eyes, nose, mouth, etc.) of fifteen groups dominate expression muscle, not only in macro(centimeter)-scale but in micro-scale. The study will provide technical supports for mechanical behavior of human face skin and expression mechanism.

作为个体身份最具鉴别性的人脸，6种典型的大尺度表情表达（惊讶、厌恶、愤怒、恐惧、悲伤、愉悦等）可以经视觉观察得到定性分析，但是皮肤变形的定量分析和细微的和瞬时的面部变化正成为一个非常重要的研究课题。问题的难点是需要开发人脸皮肤力学性能的原位测量技术和非接触式高分辨率、时序变形分析技术。本项目将开发人脸皮肤原位力学性能测量装置。在心理学专家的协助下测量和研究真人表情的实时变化。将实验测试和有限元方法相结合研究脸部皮肤的力学本构和变形传递机制。开发新的位移函数和算法，通过多相机自适应3D-DIC并行算法，实现对面部变形的区域划分和实时跟踪，实现跨尺度三维变形的精确测量。通过处理变形数据实现面部表情特征的自动识别，进而获得人脸15组主导表情肌肉的宏观（厘米级）和微观（微米级）的结构信息和特征信息（含眼、鼻、嘴等结构形状和大小），为揭示人脸的力学行为和表情形成机理提供技术支持。

人体皮肤力学性能的研究和人脸表情机理研究对于多领域都有重要的意义。本项目完成的主要研究内容是：通过人脸软物质-橡胶材料力学特性表征实验研究、平面圆膜受侧向压力起鼓的实验研究、球面压痕在一般各向异性弹性半无限体上的接触问题研究等，开展了人脸本构研究。测量和研究6种典型的大尺度表情表达（惊讶、厌恶、愤怒、恐惧 、悲伤、愉悦等）表情的实时变化。将实验测试和有限元方法相结合研究脸部皮肤的力学本构和变形传递机制。开了一套全新的透明压痕装置，采用一种特殊的光学系统实时捕捉接触图像。这套共轴光学系统由一复合物镜和一个特殊的物镜组成，后者既充当压头，也是物镜组的一部分。 通过光学设计考虑了曲面修正，保证曲面物体通过光学系统后成像在CCD面靶面上。研发了双数字图像相关（DIC）系统，实验研究了单轴拉伸下，宏观应变沿膜基界面的传递机制及薄膜表面曲率演变过程。将三维数字图像相关方法用于活体皮肤表面的变形测量，实时研究表情的发生和消失以及微表情问题。我们将主应变方向场与解剖模型做了对比，我们发现面部皮肤的主应变方向与面部肌肉模型基本保持一致，而且同一块肌肉上的主应变方向基本是相同的。这为进一步研究表情的传递机制、肌肉运动和变形传递研提供相应的实验基础。采用CT进行实验，得到头部切片。然后用Mimics软件分别提取头部骨骼和软组织部分，为了便于进一步计算，我们选取重点区域。然后我们针对此区域进行网格划分，并对加载点附近网格进行局部加密。我们采用压痕实验及双层体系模型得到弹性模量，将此参数作为参数软组织的材料参数，采用Abaqus有限元分析软件进行计算分析，然后进一步和实验数据进行对比。通过对比实验、理论以及有限元的结果，我们发现基于CT的面部软组织力学性能的有限元模型能够更好的模拟真实条件，与实验结果相符。采用直接诱发情绪状态，通过高光谱成像技术记录被试者整个面部响应。我们可以得到不同波段平静和开心状态下面部不同位置的光相应信息。