一种基于正面头像的三维人脸建模方法

本文提出了一种真实感三维人脸自动建模方法,只需一张光照均匀的人脸正面头像,通过自动提取面部的关键特征点重建人脸的几何模型,最后从图像上获取面部纹理信息得到真实感的三维人脸。实验表明该方法能较真实有效地重建三维人脸。     

基于单张正面人脸图像的三维人脸自动建模系统实现

研究了一种基于正面人脸照片的真实感三维人脸自动重建方法,并运用计算机视觉图OpenCV和图形开发库OpenGL,在VC++6.0环境下开发了三维人脸自动建模系统。该系统对输入的人脸照片首先进行人脸检测,在检测到的区域进行人脸关键特征提取,并根据检测到的特征点的几何信息对CANDIDE-3模型进行整体和局部调整,得到个性化的三维几何人脸,最后从人脸图像上获取面部纹理信息并得到真实感的三维人脸。     

基于单幅图像的人脸三维建模研究

由于人脸三维模型在三维动画、计算机游戏、视频会议、医学手术以及生物教学等许多领域都有着广泛的应用价值,使得越来越多的研究人员开始思考获取人脸三维模型的有效途径。文中先采用AdaBoost算法进行人脸检测,判断输入图像中是否含有人脸;接着,采用ASM模型对含有人脸的图像进行特征点提取;最后依据从图像中提取的特征点等信息对初始模型进行调整,实现了一种基于单幅图像的建模方法。     

真实感三维人脸建模技术综述

三维建模是计算机图形学和计算机视觉领域的基本问题。人脸具有共性强及个性鲜明的特点,成为众多三维建模算法的实验平台。但由于人脸的多样性、复杂性,建立真实感强的目标三维人脸模型在学术研究和实际应用方面都具有重要意义。在文献梳理的基础上,阐述了真实感三维人脸建模理论和实践相关研究成果,总结了目前三维人脸重建的研究方法,分析了各种三维人脸重建方法的优缺点,提出了有待进一步研究的问题,展望了未来的发展趋势。     

基于彩色照片的人脸三维建模方法

提出了一个基于彩色照片的个性化人脸建模的新方法。根据MPEG-4确定出标准模型特征点的位置,从正面和侧面两幅照片出发,首先进行了这两幅彩色图像人脸边缘特征提取,接着定位进行特征点编辑,获取人脸的关键特征点的位置,然后对标有对应特征点的标准模型进行变形,进行纹理映射,最后获得了真实的个性化人脸模型,两张照片可以合成任意角度视图。以该方法建立的人脸模型精度比较高,可以快速地建立个性化人脸模型,为三维人脸动画提供了真实模型。     

基于彩色照片的人脸三维建模方法

提出了一个基于彩色照片的个性化人脸建模的新方法.根据MPEG-4确定出标准模型特征点的位置.从正面和侧面两幅照片出发,首先进行了这两幅彩色图像人脸边缘特征提取,接着定位进行特征点编辑,获取人脸的关键特征点的位置,然后对标有对应特征点的标准模型进行变形,进行纹理映射,最后获得了真实的个性化人脸模型,两张照片可以合成任意角度视图.以该方法建立的人脸模型精度比较高,可以快速地建立个性化人脸模型,为三维人脸动画提供了真实模型.     

基于多幅照片的人脸三维建模

在通用三维人脸模型的基础上，提出了一种基于多幅不同角度的人脸照片的三维人脸建模方法。该方法从多幅照片中提取面部特征点，修改通用人脸模型来生成具体的人脸三维模型。     

真实感人脸重建技术

真实感人脸的建模,面临着人脸生理结构复杂,提取人脸的图像或者视频易受光照等环境影响,表情非常丰富等困难,因此在计算机图形学领域,吸引了许多学者进行研究。本文对近年来真实感人脸建模方面的研究进行了详细的综述,并对其未来的研究方向提出一些建议。     

改进的曲面人脸模型纹理映射技术

基于Bezier曲线进行人脸建模,并实现了从一般人脸模型到特定人脸模型修改。为了使特定的人脸模型更加具有真实感,必须给这个曲面人脸模型赋予真实的纹理信息。该文引入多视角图像纹理信息技术,取得很好的效果。     

基于投影与深度学习网络的三维人脸特征点定位方法

标定三维人脸模型特征点对人脸识别、人脸建模等都具有重要作用。针对人脸特征点标定需要手工干预、标定特征点个数少或不准确、标定时间长等问题,提出了一种基于投影与深度学习网络的人脸三维模型特征点标定法。基于正交投影,生成人脸三维模型二维深度图与二维特征点位置,采用以卷积神经网络为主的深度学习网络模型训练测试,将深度图上特征点映射到三维人脸模型,实现眉毛、眼睛、鼻尖、嘴巴等重要区域的特征点定位。实验表明,该方法可自动标定三维人脸模型特征点,快速、准确预测足够数量特征点位置。     

Sample based 3D face reconstruction from a single frontal image by adaptive locally linear embedding

In this paper, we propose a highly automatic approach for 3D photorealistic face reconstruction from a single frontal image. The key point of our work is the implementation of adaptive manifold learning approach. Beforehand, an active appearance model (AAM) is trained for automatic feature extraction and adaptive locally linear embedding (ALLE) algorithm is utilized to reduce the dimensionality of the 3D database. Then, given an input frontal face image, the corresponding weights between 3D samples and the image are synthesized adaptively according to the AAM selected facial features. Finally, geometry reconstruction is achieved by linear weighted combination of adaptively selected samples. Radial basis function (RBF) is adopted to map facial texture from the frontal image to the reconstructed face geometry. The texture of invisible regions between the face and the ears is interpolated by sampling from the frontal image. This approach has several advantages: (1) Only a single frontal face image is needed for highly automatic face reconstruction; (2) Compared with former works, our reconstruction approach provides higher accuracy; (3) Constraint based RBF texture mapping provides natural appearance for reconstructed face.     

一种三维人脸表情动画快速生成方法

面向动画创作的三维人脸表情动画生成框架由人脸模型简化、特征点驱动以及表情动画生成等三部分组成。以MPEG-4人脸动画定义标准为基础,提出了以脸部定义参数流驱动关键特征点的动画生成算法。实验结果表明：该方法实现了真实感和实时性的有效结合,能满足动画创作的要求。     

基于特征点配准形变的三维人脸表情合成

针对视频监控中被部分遮掩脸部图像的还原不准问题,提出了特征点配准形变的三维人脸表情合成算法.将目标人脸图像灰度化后建立三角人脸网络,提取人脸未遮掩部分特征点,与通用模型特征点配准,形变而合成新的网络人脸模型.从正面及侧面分别进行表情纹理映射,根据光照参数、老化参数合成三维正面及侧面人脸表情纹理图像.实验证明,提出的算法相对于同类合成算法,时间复杂度较低,精确度较高,鲁棒性较强,且随特征点被遮掩程度的增大,其优势更明显.     

基于矩－傅里叶描述子的不同姿态三维人脸识别

三维人脸相比二维人脸包含更丰富的信息,而且能够克服姿态,表情,光照因素的影响,更好地表示人脸特征,从而逐渐受到广泛的关注和重视。文中对不同姿态下的三维人脸深度图,先用微分几何相关知识把该图校正到正中面,然后通过提取人脸面部的等高线特征,将三维人脸变成容易处理的二维曲线,针对如何能够更好的描述该二维曲线,提出了一种把矩和傅里叶描述子相结合的新方法,最后利用提取的曲线特征进行人脸识别。实验结果表明,该方法具有很好的实时性和更强大的形状区分能力,鲁棒性高,优于传统傅里叶描述子提取等高线特征的方法。     

Kernel Model Applied in Kernel Direct Discriminant Analysis for the Recognition of Face with Nonlinear Variations

A kernel-based discriminant analysis method called kernel direct discriminant analysis is employed, which combines the merit of direct linear discriminant analysis with that of kernel trick. In order to demonstrate its better robustness to the complex and nonlinear variations of real face images , such as illumination, facial expression, scale and pose variations, experiments are carried out on the Olivetti Research Laboratory, Yale and self-built face databases. The results indicate that in contrast to kernel principal component analysis and kernel linear discriminant analysis, the method can achieve lower (7%) error rate using only a very small set of features. Furthermore, a new corrected kernel model is proposed to improve the recognition performance. Experimental results confirm its superiority (1% in terms of recognition rate) to other polynomial kernel models.     

Automatic location of main facial features in front-view images

Ｆｒｏｎｔ－Ｖｉｅｗ ＩｍａｇｅｓTX１ＩｎｔｒｏｄｕｃｔｉｏｎＤｕｅｔｏｔｈｅｉｎｃｒｅａｓｉｎｇｄｅｍａｎｄｉｎｐｅｒｓｏｎａｌｉｄｅｎｔｉｆｉ－ｃａｔｉｏｎａｎｄｓｅｃｕｒｉｔｙｗｏｒｋ,ａｕｔｏｍａｔｉｃｆａｃｅｒｅｃｏｇｎｉｔｉｏｎｈａｓｂｅ...     

When Adolescents Represent the Third Dimension: Three Case Studies

The main focus of this article is the representation of the third dimension. The sample is sixty adolescent 14‐year‐olds. Our research is concerned with the study of the drawings of the same array of 3D objects in three cases: the representation of 3D objects without the presence of models (through verbal instructions), by observation of physical models, and by observation of their digital models on a computer screen. The study examines the drawing techniques applied by adolescents to represent the 3rd dimension and the impact of the model (3D physical or digital model) in the drawing outcome. A comparison study of the three kinds of drawings of the same array of objects showed that 14‐year‐old children face serious difficulties in depicting the third dimension in the absence of a model, using mainly a mixture of drawing devices typical of earlier stages, according to theories of drawing development. On the other hand, when drawing from observation, the nature of the model seems to play an important role in the drawing outcome, with a clear superiority in performance when the model is digital. The findings of this study suggest that the more extended use of 3D models, either physical or digital, could help the better understanding of spatial relationships and evoke the use of more advanced drawing techniques for the depiction of 3D layouts. The study also suggests that a better understanding of the nature of graphic development in adolescence is essential for many areas of the curriculum.