摘要: 已有人脸表情特征提取算法的表情识别率较低。为此,提出一种基于链码的人脸表情几何特征提取算法。以主动形状模型特征点定位为基础,对面部目标上定位的特征点位置进行循环链码编码,以提取出人脸表情几何特征。实验结果表明,相比经典的LBP表情特征鉴别方法,该算法的识别率提高约10%。
关键词:
链码,
主动形状模型,
面部目标,
几何特征,
表情特征,
表情识别
Abstract: The existing facial expression recognition rate of facial expression feature extraction algorithm is low. For this, this paper proposes a facial geometric feature extraction algorithm based chain codes. Based on active shape model that locates feature points and outputs the points’ coordinates of facial targets the coding method gives a circular codes to extract the facial geometric feature. Experimental results show that, compared with the method of typical LBP expression recognition, the accuracy of the algorithm is increased by nearly 10%.
Key words:
chain code,
Active Shape Model(ASM),
facial target,
geometrical feature,
expression feature,
expression recognition
中图分类号:
张庆, 代锐, 朱雪莹, 韦穗. 基于链码的人脸表情几何特征提取[J]. 计算机工程, 2012, 38(20): 156-159.
ZHANG Qiang, DAI Dui, SHU Xue-Ying, HUI Sui. Facial Expression Geometrical Feature Extraction Based on Chain Code[J]. Computer Engineering, 2012, 38(20): 156-159.