一种高性能部分遮挡的人脸表情识别方法

A Novel High-quality Partially Occluded Face Expression Recognition Method

针对部分遮挡的人脸表情识别,提出了一种基于部分遮挡判断的人脸表情识别方法。首先对包含表情信息的静态图像进行Gabor小波变换,提取表情特征矢量,然后用最大相关分类器对特征矢量分类。实验表明,该方法对局部遮挡的人脸表情识别有较强的鲁棒性,对无遮挡的识别也有所改善。

This paper presented a partially occluded face expression recognition method with sub-region occlusion judgment.Given a still image containing facial expression information,expression feature vectors are extracted by Gabor wavelet transformation firstly.Secondly,the feature vectors are classified by using a maximum correlation classifier.Experiment shows that expression the facial expression recognition method provides robustness against partial occlusion and non-occluded faces.