Computer Engineering and Applications ›› 2014, Vol. 50 ›› Issue (20): 149-157.

Previous Articles     Next Articles

Summary of facial expression recognition

WANG Dawei, ZHOU Jun, MEI Hongyan, ZHANG Su’e   

  1. School of Electronics & Information Engineering, Liaoning University of Technology, Jinzhou, Liaoning 121001, China
  • Online:2014-10-15 Published:2014-10-28

人脸表情识别综述

王大伟,周  军,梅红岩,张素娥   

  1. 辽宁工业大学 电子与信息工程学院,辽宁 锦州 121001

Abstract: As a research direction of the affective computing, facial expression recognition which constitutes the basis of emotion understanding, is the premise to complete human-computer interaction intelligent. Facial expression is so exquisite that it consumes a large amount of computation time and influences the timeliness and experience feeling from human-computer interaction intelligent. Consequently, facial feature extraction has become an important research topic in the area of facial expression recognition. The progress and stable framework for facial expression recognition in recent five years are generalized, a serial of algorithms applied in feature extraction and expression classification are summarized, Then, the main algorithms and their improvement are described in detail, and advantages and disadvantages among in different algorithms are analyzed and compared. In the same time, comparison with the other algorithms is also introduced. The challenges and shortcomings are pointed out by the research of practical problems in facial expression recognition application.

Key words: expression recognition, feature extraction, feature classification, algorithm improvement, algorithm comparison

摘要: 人脸表情识别作为情感计算的一个研究方向,构成了情感理解的基础,是实现人机交互智能的前提。人脸表情的极度细腻化消耗了大量的计算时间,影响了人机交互的时效性和体验感,所以人脸表情特征提取成为人脸表情识别的重要研究课题。总结了国内外近五年的人脸表情识别的稳固框架和新进展,主要针对人脸表情特征提取和表情分类方法进行了归纳,详细介绍了这两方面的主要算法及改进,并分析比较了各种算法的优势与不足。通过对国内外人脸表情识别应用中实际问题进行研究,给出了人脸表情识别方面仍然存在的挑战及不足。

关键词: 表情识别, 特征提取, 表情分类, 算法改进, 算法对比