基于Eigenspace的面瘫判定系统  

A FACIAL PARALYSIS JUDGING SYSTEM BASED ON EIGENSPACE

作　　者：曾文珺[1] 戚飞虎[1] 陶峻[1] 赵燕玲[2] 

机构地区：[1]上海交通大学计算机科学与工程系计算机视觉实验室,上海200030 [2]上海复旦大学附属五官科医院,上海200031

出　　处：《计算机应用与软件》2003年第12期70-72,共3页Computer Applications and Software

摘　　要：本文将特征空间分类方法应用于面瘫图像的判定。结合应用的特点 ,引入多特征空间分类 ,比较图像与它们到两个特征空间的投影之间的距离作出判定 ,并在系统中加入灰度图变换 ,脸部子区域分割 ,确定对称轴和输入向量计算等前处理步骤。实验证明 ,本文的改进提高了系统判定的正确性。In this paper,Eigenspace classification method is applied to facial paralysis judgement.Considering the peculiarity of the application,multi-eigenspace classificaion is introduced into the system and images are judged by the likelihood(eigen distance)between the vectors and their projections on the eigenspaces.And the paper presents some pre-process including the translation of the gray images,the segmentation of sub-region,the decision of symmetric axis and the calculation of input vectors.The experiments demonstrated that the improvements enhance the correctness of the system.

关 键 词：面瘫判定系统 EIGENSPACE 计算机视觉 特征空间分类方法 医院 主分量分析 图像特征 

分 类 号：R745.12[医药卫生—神经病学与精神病学] TP391.41[医药卫生—临床医学]