红外面部图像的贝尔面瘫针灸穴位定位算法  

Acupuncture Points Localization Algorithm About Bell′s Palsy on Facial Infrared Images

作　　者：赵阳[1] 张涛[1,2] 练秋生[1] 

机构地区：[1]燕山大学信息科学与工程学院,河北秦皇岛066004 [2]燕山大学电气工程学院,河北秦皇岛066004

出　　处：《小型微型计算机系统》2012年第7期1613-1619,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61071200;60904100)资助;中医药行业科研专项(200707022)资助;河北省自然科学基金项目(F2011203073)资助

摘　　要：将红外图像处理应用于中医诊疗,提出一种面部穴位自动定位算法,为贝尔面瘫的针灸选穴提供了客观化和标准化的依据.首先利用头部阈值区分前景(即头部图像)和背景,在前景中根据"三庭五眼"规则对面部特征进行分区;采用Mini-mumEigenvalue算子检测面部特征的角点,Log算子检测面部特征的边缘,综合应用角点和边缘信息,定位各面部特征的位置;最后,以面部特征的位置作为参考坐标,完成面部穴位的定位.经过大量面部红外图像验证,本方法定位平均准确率达94%,其中人中穴的定位准确率接近100%.This paper uses infrared image processing into traditional Chinese medicine diagnosis,put forward a kind of positioning algorithm to automatic positioning facial acupuncture points,and provide an objective and standardization basis for Bell ′s palsy to choose acupuncture.This method firstly distinguishes the foreground（namely head images） and background use head threshold,divide the facial features in accordance with the rule of ＂three court five eyes＂.Then use MinimumEigenvalue operator detect facial features′ corner,use Log operator detect facial features′ edges,integrated apply the corner and edge information,positioning the position of each facial feature;Finally,use the facial features position as a reference coordinate,complete facial acupuncture points positioning.After the validation of many facial infrared images,this method′s average accuracy is over 94%,the accuracy of ren zhong is close to 100%.

关 键 词：红外图像 穴位定位 贝尔面瘫 角点检测 边缘检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]