【摘要】 人脸表情是人与人之间进行信息交流的一种重要的信息传递媒介,自上世纪70年代,自动人脸表情特征分析驱动着大量的研究,尤其是在人脸表情识别这一领域。相比于之前采用传统图像处理方式先进行特征提取再结合各类分类器进行人脸表情识别,结合近几年发展迅速的深度学习技术,提出一种基于深度学习下的人脸表情特征分析,人脸图像有着特有的特性,在进行网络训练前期会进行图像预处理,然后再进行CNN的网络训练,实验数据主要基于JAFFE+人脸表情库中7中表情包括(自然、生气、恐惧、厌恶、高兴、惊讶、悲伤)进行实验来验证深度学习下人脸表情识别的有效性,体现在高的识别率和易操作性。更多还原

【Abstract】 Facial expression is an important medium of information transmission between people and people, since the 70 s of the last century, automat-ic facial expression analysis drives a lot of research, especially in the field of facial expression recognition. Compared with the previous tra-ditional image processing methods, the feature extraction is combined with various classifiers for facial expression recognition, with deeplearning techniques with rapid development in recent years, proposes an analysis of facial expression features based on depth learning.Face images have unique features, image pre-processing is carried out in the early stage of network training, and then do CNN training, theexperimental data are conducted to validate the effectiveness of deep learning mainly based on JAFFE face database and FER2013 data-base(including seven expressions of nature, anger, fear, disgust, joy, surprise, sadness), finally high recognition rate and maneuverabilityare shown.更多还原