结合FDM与STLBP-IP特征的微表情识别

Micro-expression Recognition Combined with FDM and STLBP-IP Features

针对面部动力谱(FDM)特征易受光照影响、对运动信息描述不准确的缺陷,提出基于FDM特征与时空局部二值模式积分投影(STLBP-IP)特征相结合的微表情识别方法。将FDM特征与STLBP-IP特征相结合,在弥补FDM对运动信息描述不足的同时对人脸面部信息进行补充描述以提升识别率。使用支持向量机进行分类,在SMIC和CASMEⅡ微表情数据库上进行实验。实验结果表明,该算法识别率有所提高,分别达到57.14%和64.59%。

Focusing on the problem that the facial dynamics map(FDM)feature is susceptible to the influence of light,and thus motion information tracking is inaccurate,a fusion method combining FDM with spatiotemporal local binary pattern with integral projection(STLBP-IP)is proposed for micro-expression recognition.This paper combines FDM features with STLBP-IP features to make up for the lack of FDM's description of motion information,while also supplementing the description of facial information to improve the rec⁃ognition rate.Adopting a support vector machine as classifier,and experiments on SMIC and CASMEⅡmicro-expression database.The experimental results show that the recognition rate of the algorithm discussed in this paper has improved,reaching 57.14%and 64.59%respectively.