权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Development of Three-Dimension computer graphic system for reconstruction of the face features from the skull.

开发三维计算机图形系统，用于从头骨重建面部特征。

基本信息

批准号：
07557269
负责人：
YUASA Kenji
金额：
$ 1.22万
依托单位：
KYUSHU UNIVERSITY
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (A)
财政年份：
1995
资助国家：
日本
起止时间：
1995 至 1997
项目状态：
已结题

项目摘要

At first We analyzed the relationships between the size of organs that formed face features such as the masseter muscle and the parotid gland, and characteristic of each subject such as sex, age, the size of the skull, height, weight, body mass index and number of residual molar teeth. The size of each organs were measured using CT images. We evaluated the relationships using multiple regression analysis. And the function for predict value of each organ in face was made. Their multiple correlation coefficients were 0.71 to 0.74.We evaluated the grade of reconstruction of face that persons could identify each subject using cropped faces of each subject which were reconstructed using the eigen-face methods, which is based on a methods known as the principal component analysis. Furthermore, the facial images was classified by cluster analysis. As a results, It was consider that individual face was identified to cropped face with at least 7 eigen-faces, that is, the grade of reconstruction was 0.88 of correlation coefficients. Furthermore, face images were classified into 5 groups by cluster analysis.We analyzed the factors in face by which persons identify individuals. As results, The most important factor was the distance between both eyes. Secondly, the width of the lips.As above mentioned, we obtained useful results concerning with the relationships between face and the individual features (skull size, sex, age, body mass, masticator function), and the grade of reconstruction of face for identifying the individual face. However, we are developing the computer graphic software for reconstruction of the face features now.

首先分析了咬肌、腮腺等构成面部特征的器官的大小与性别、年龄、颅骨大小、身高、体重、体重指数、剩余磨牙数等个体特征之间的关系。使用CT图像测量每个器官的大小。我们使用多元回归分析来评估这些关系。并建立了面部各器官预测值的函数。它们的复相关系数为0.71 ~ 0.74。我们用基于主成分分析的特征脸方法重建的每个受试者的裁剪脸来评价人们能够识别每个受试者的脸的重建等级。最后，利用聚类分析对人脸图像进行分类。实验结果表明，该方法可以将人脸识别为具有至少7个特征脸的裁剪人脸，即重建的相关系数为0.88。通过聚类分析，将人脸图像分为5类，并分析了人脸图像中人们识别个体的因素。结果表明，影响视力的最重要因素是双眼间距。第二，唇宽，如上所述，我们得到了有关面部与个体特征（颅骨大小、性别、年龄、体重、咀嚼肌功能）之间的关系以及面部重建等级的有用结果，以识别个体面部。然而，我们正在开发的计算机图形软件重建的人脸特征。