基于唇色信息的唇部检测方法研究  

Research of Lip Detection Based on Lip Color Information

作　　者：杨冬涛[1] 罗聪[1] 龚昌来[1] 

机构地区：[1]嘉应学院电子信息工程学院,广东梅州514015

出　　处：《嘉应学院学报》2010年第11期35-38,共4页Journal of Jiaying University

摘　　要：唇部检测是唇读识别的基本步骤,在研究人类视觉和唇色聚类的基础上,分析了唇色和肤色的聚类性,提出了利用神经网络单层感知器学习算法对彩色人脸图像的唇部进行定位的新方法。实验表明,采用感知器学习算法比F isher方法的唇部定位更准确,轮廓更清晰,更有利于唇部口型的识别。Lip detection is the basic step of lip recognition.Based on the research of human vision and lip color clustering features and the analysis of clustering between lip color and skin color,a new method using neural network single perception learning algorithm for lip positioning of color facial image has been proposed.Experiments show that using perception approach for lip detection is more accurate and clearer than Fisher approach,which is more conductive to lip recognition.

关 键 词：唇读 唇部定位 YIQ 感知器算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]