【摘要】 表情识别是基于视觉信息将脸部的运动或脸部特征的形变进行分类,包括三部分:脸部定位、脸部特征抽取和表情分类.本文首先使用肤色模型进行脸部定位;对提取出来的人脸进行预处理,然后通过Canny算子和人脸形状模型相结合的Can-ny-AAM方法进行特征点定位;最后利用曲线拟合的方法进行特征提取.基于上述算法建立表情识别平台,经过大样本对实时表情识别验证,结果表明对于不同光照下的实时表情识别具有鲁棒性.更多还原

【Abstract】 Facial expression recognition deals with the classification of facial motion and facial feature deformation into abstract classes that are purely based on visual information.It is mainly about the location of the face,the extract of the face feature,and the classification of the expression.Image preprocessing is to eliminate the interference of the illumination and scale.Firstly,skin tone model is used to locate the face area,then preprocessing the extract image.Then combine the operator of Canny with Active Appearance Model(Canny-AAM) to location the feature point,and extract the face feature by regression lastly.The facial expression recognition system based on the algorithm of this paper was setup.Large real time facial expression recognition in different illumination environment samples show that the system is robust.更多还原