摘要
为了提高面部表情的识别性能,文章提出了基于PLBP(金字塔LBP)的表情识别算法。该方法即通过多尺度分析来建立人脸图像的局部二值模式(Local Binary Pattern,LBP)金字塔特征,能有效地提取人脸图像的全局和局部特征。表情识别系统的整个流程包括人脸特征的点跟踪、人脸的面部区域划分、PLBP特征提取和情感分类。文章在CK+表情数据库上进行了对比实验,实验结果表明,PLBP特征提取在面部表情识别中具有较高的识别率。这体现了全局和局部特征相结合对于图像识别的重要性。
In order to improve the performance of face expression recognition, we propose a algorithm for expression recognition based on pyramid local binary pattern (PLBP) feature. This method means that the face image pyramid LBP is constructed by the multi-scale analysis. It can effectively extract the global and local features. The whole process of the recognition system is including the face feature points tracking, the face division, the feature extraction of PLBP and the emotion classification. The contrast experiment is carried out in the CK+ database. The experimental results demonstrate that the extraction of PLBP feature achieves high recognition rates in the face expression recognition, which reflects the combination of global and local features is very significant to the image recognition.
出处
《信息化研究》
2016年第2期47-50,共4页
INFORMATIZATION RESEARCH
关键词
金字塔LBP
多尺度分析
表情识别
全局和局部特征
pyramid local binary pattern
multi-scale analysis
expression recognition
global and local features