基于协作低秩分层稀疏和LC-KSVD的人脸表情识别

为了降低人脸表情识别对待识别个体的依赖程度,控制识别字典规模,增加识别准确度,提出了一种基于协作低秩和分层稀疏的表情识别字典构建方法.通过协作低秩和分层稀疏表示(C-HiSLR)有效分离与待识别个体相关部分,保留表情变化部分,并结合标签一致区分字典学习(LC-KSVD)算法,进行相应待训练表情序列的重构识别和对应类别字典的区分程度的优化学习.该方法在CK+数据集上进行验证,识别效果较一般基于稀疏表示模型算法有明显的提升.

Facial expression recognition based on collaborative low-rank and hierarchical sparse and LC-KSVD

To reduce dependence of individual on facial expression recognition,control scale of recognition dictionary,increase recognition accuracy,propose a dictionary recognition construction method based on collaborative low-rank and hierarchical sparse representation. Effective separation the relevant characteristics of the individual to be identified by collaboration lowvrank and hierarchical sparse model (C-HiSLR ),keep expression variation characteristics,combined with label consist KSVD(LC-KSVD),to identify and reconstruct the sequence of expression to be trained and optimization study of distinction between corresponding category dictionary. The methods is verified on data set in CK+,recognition effect is significantly improved than general model algorithm based on sparse representation.