人脸表情识别研究进展  

Research progress of facial expression recognition

作　　者：姜月武 路东生 党良慧 杨永兆 施建新 JIANG Yuewu;LU Dongsheng;DANG Lianghui;YANG Yongzhao;SHI Jianxin(School of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学电子电气工程学院,上海201620

出　　处：《智能计算机与应用》2021年第6期43-50,共8页Intelligent Computer and Applications

基　　金：上海市自然科学基金(17ZR1411900);上海市科委重点项目(18511101600)。

摘　　要：随着计算机视觉的发展和人工智能产业的兴起,人脸表情识别技术在人工智能产业有着广泛的应用需求。人脸表情识别在传统机器学习算法下对环境及姿态的改变不具备良好的鲁棒性,而且识别精度也达不到实际应用的要求。计算机和图像处理器等硬件性能的提升,以大数据为核心的深度学习算法得到快速发展,人脸表情识别技术开始趋于在深度学习算法上研究。本文分别对人脸表情图像预处理、特征提取、特征分类3个关键技术进行介绍具体叙述了从传统的机器学习到基于深度学习的人脸表情识别技术的研究进展,分析了人脸表情识别技术目前面临的挑战和发展趋势。With the development of computer vision and the rise of artificial intelligence,facial expression recognition(FER)technology has a wide range of applications in the artificial intelligence industry.FER under the traditional machine learning algorithm does not have good robustness to the change of environment and posture.Moreover,the accuracy of identification is not up to the requirement of practical application.With the improvement of hardware performance such as GPU,deep learning algorithm with big data as its core has been developed rapidly.FER technology tends to be studied in deep learning algorithm.This paper introduces three key technologies of facial expression image preprocessing,feature extraction and feature classification.The research progress from traditional machine learning to FER based on deep learning is described,and the current challenges and development trend of facial expression recognition technology are analyzed.

关 键 词：人脸表情识别 深度学习 特征提取 特征分类 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]