Computer Engineering and Applications ›› 2023, Vol. 59 ›› Issue (23): 28-47.DOI: 10.3778/j.issn.1002-8331.2303-0401

• Research Hotspots and Reviews • Previous Articles     Next Articles

Research Progress in Three-Dimensional Face Recognition

LIU Li, GONG Yong, ZHAO Guoqiang   

  1. 1.China Nanhu Academy of Electronics and Information Technology, Jiaxing, Zhejiang 314002, China
    2.College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou 310027, China
  • Online:2023-12-01 Published:2023-12-01

三维人脸识别研究进展

刘力,龚勇,赵国强   

  1. 1.中国电子科技南湖研究院,浙江 嘉兴 314002
    2.浙江大学 信息与电子工程学院,杭州 310027

Abstract: Three-dimensional face recognition technology has unique advantages such as natural, convenient and user-friendly, and therefore has become a trending research direction in both academia and industry. In recent years, researchers have proposed various new methods to challenge 3D face recognition for different types of 3D face data such as monocular RGB color images, RGB-depth images, point cloud data, and multi-modal data, etc. According to the different formats of 3D face data, 3D face recognition methods are classified into 3D face reconstruction methods from monocular RGB images, face recognition methods based on point cloud data, RGB-depth image data, and multi-modal data. The principles, differences, and connections of various methods have been compared and analyzed. The latest progress of 3D face recognition research is comprehensively summarized, and the advantages and disadvantages of various technologies are expounded. Face databases in different formats are introduced, analyzing data collection methods, expression changes, posture changes. Finally, challenges faced by 3D face recognition technology and its future development direction are discussed.

Key words: 3D face, face recognition, feature extraction, 3D face database

摘要: 三维人脸识别技术具有自然、便捷、用户体验友好等独特优势,已成为学术界和工业界的一个热门研究方向。近几年,针对单目彩色图像、彩色图-深度图数据、点云数据、多模态数据等不同类型的三维人脸数据,研究者们提出了各种新方法来挑战三维人脸识别。根据三维人脸数据的不同格式,将三维人脸识别方法分为从单目彩色图像进行三维人脸重构的识别方法、基于点云数据的人脸识别方法、基于彩色图-深度图数据的人脸识别方法,以及基于多模态数据的人脸识别方法,对比分析了各类方法的原理、区别和联系,全面总结了三维人脸识别研究的最新进展,阐述了各类技术的优势和弊端。介绍了不同格式的人脸数据库,从数据收集方式、表情变化、姿态变化等方面进行分析,最后讨论了三维人脸识别技术面临的挑战和未来的发展方向。

关键词: 三维人脸, 人脸识别, 特征提取, 三维人脸数据库