欢迎访问《图学学报》 分享到:

图学学报

• 计算机图形学 • 上一篇    下一篇

基于 Kinect 的拉普拉斯网格形变三维人脸建模

  

  1. 1. 河南理工大学计算机科学与技术学院,河南 焦作 454000; 
    2. 广东星辉天拓互动娱乐有限公司(杭州分公司),浙江 杭州 310000
  • 出版日期:2018-10-31 发布日期:2018-11-16
  • 基金资助:
    国家自然基金青年基金项目(61503124);河南省科技攻关项目(162102310090,172102210273);河南省教育教学改革项目(2014SJGLX478)

Laplace’s Grid Deformation 3D Face Modeling Based on Kinect

  1. 1. College of Computer Science and Technology, Henan Polytechnic University, Jiaozuo Henan 454000, China;
    2. Guangdong Xinghui Tiantuo Interactive Entertainment Corporation, Hangzhou Zhejiang 310000, China
  • Online:2018-10-31 Published:2018-11-16

摘要: 为了快速创建真实感较强的三维人脸模型,提出了基于 Kinect 的拉普拉斯网格形 变建模方法。利用 Kinect 获取彩色和深度图像信息,对深度图像进行双边滤波处理,对彩色图 像进行低层级顶点定位;构建标准三维人脸模型,并为该人脸模型中的顶点建立低、中、高 3 个级别的层级结构,通过低、中层级中顶点的位置关系创建 Sibson 局部坐标约束;利用该约束 构建彩色图像中间层级顶点,并结合深度信息对标准三维人脸模型进行拉普拉斯网格变形,获 得真实感较强的三维人脸模型。实验结果表明,该算法在建模的真实感上得到了提高,与对比 算法相比,在建模时间上得到很大的优化。

关键词: Kinect, 拉普拉斯网格形变, 层级模型, 人脸建模

Abstract: In order to create a realistic 3D face model in a fast speed, a Laplacian mesh deformation modeling method based on Kinect is proposed. The Kinect is used to obtain color and depth image information, perform bilateral filtering on the depth image, perform low-level vertex positioning on the color image, construct a standard 3D face model, and establish low, middle, and high levels for the vertices in the face model. In the hierarchical structure, Sibson local coordinate constraints are created through the positional relationship of the vertices in the low and middle levels; this constraint is used to construct the middle level vertices of the color image and the Laplacian mesh deformation of the standard 3D face model is combined with the depth information to obtain a realistic 3D face model. The experimental results show that the proposed algorithm is improved in the realism of modeling, and compared with the comparative algorithms, the length of modeling time is greatly optimized.

Key words: Kinect, Laplace’s grid deformation, hierarchical model, face modeling