光学技术, 2018, 44 (1): 25, 网络出版: 2018-02-01  

基于位移特征与随机森林的表情识别方法研究

Facial expression recognition based on displacement feature and random forest
作者单位
1 暨南大学 理工学院, 广州 510632
2 暨南大学 电气自动化研究所, 广东 珠海 519070
摘要
为了有效地对图像序列进行面部表情识别, 提出一种基于主动形状模型(active shape model, ASM)结合Lucas-Kanade(LK)光流法的方法提取位移特征, 采用随机森林分类器对提取到的位移特征进行分类与识别。在Extended Cohn-Kanade(CK+)人脸表情数据库上的实验表明, 该特征提取方法能够很好地描述图像序列中所包含的表情信息和特征点运动变化信息, 比常用的K-近邻、贝叶斯网络和支持向量机等分类器所表现出来的效果要好, 其识别率达到95.1%。
Abstract
In order to effectively recognize facial expression from image sequences, a new method is proposed which combines active shape model (ASM) with Lucas-Kanade (LK) optical flow to extract displacement feature, and random Forest classifier is applied for classifying and recognizing the displacement feature which effectively improves the classification results in facial expression recognition. The experiments on CK+ facial expression database demonstrate that the method can effectively represent image sequences’ expression information and motion information of feature points. Compared with conventional classifiers such as K-Nearest neighbors algorithm, Bayesian network and support vector machine, the higher performance can is achieved by the new method and its rate reached 95.1%.

林子澄, 黄元亮, 刘一民. 基于位移特征与随机森林的表情识别方法研究[J]. 光学技术, 2018, 44(1): 25. LIN Zicheng, HUANG Yuanliang, LIU Yimin. Facial expression recognition based on displacement feature and random forest[J]. Optical Technique, 2018, 44(1): 25.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!