基于稀疏表达和改进的LBP算子的人脸表情识别

摘要

基于局部二值模式(LBP)算子在人脸表情识别中直方图维数高、判别能力差、具有冗余信息等缺点,提出一种中心二值模式(CBP)算子并对人脸表情关键部位(眉毛、眼睛及嘴巴部分)提取特征.最后利用稀疏表达分类器对提取的表情特征进行识别.实验结果表明,该算法的识别效果有了极大的提高.%According to the drawbacks of local binary pattern ( LBP) operator that it has high histogram dimension, poor discriminate ability and redundant information in facial expression recognition, the centre binary pattern ( CBP) operator is proposed and employed to extract the feature of key parts of facial expression ( eyebrow, eye and mouth). Finally, the sparse representation classifier is used to recognise these extracted features. Experiments show that the recognition effect of the proposed algorithm is greatly improved.