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

Development of analysis of velopharyngeal movement using digital human models and application for cleft palate speech

使用数字人体模型开发腭咽运动分析及其在腭裂语音中的应用

基本信息

批准号：
18390540
负责人：
MISHIMA Katsuaki
金额：
$ 7.55万
依托单位：
Okayama University
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (B)
财政年份：
2006
资助国家：
日本
起止时间：
2006 至 2007
项目状态：
已结题

项目摘要

1. Development of motion analysisHomologous modeling, FED, and principal component analysis were applied Ear motion range images which were generated using three infrared-red cameras and a color camera with an infrared-red pattern projected on a face, and lip motion analysis of normal individuals was then performed. After principal component scores, which indicated characteristics of lip motion, were calculated at phonation of each vowel, principal component scores were plotted in the two-dimension. Discriminant analysis revealed the differences between lip motions at phonation of /u/ and /i/ could be detected. Therefore, it was suggested that principal component analysis was useful for lip motion analysis.2. Production of range images using a streoendoscopeUsing a stereoendoscope, which was 5.4mm diameter and included two telescopes, measuring system was developed. Preliminary studies disclosed the following three problems of the stereoendoscope. 1) The center of the image was deviated. 2) The distortion became big so as to leave the center of the image. 3) There was a difference between the brightness of two images. Tsai algorithm was applied to the problems of 1) and 2), LoG filtering was applied to the problem of 3). By applying process of rectifying and a Birchfield's algorithm range images could be then produced from the image obtained from the stereoendoscope successfully.3. Relationship between velopharyngeal shapes and nasalance scores using velopharyngeal cephalogramsRelationship between measurements of velopharyngeal cephalograms at blowing and phonation /a/ and nasalance scores obtained from Nasometer was investigated with a multivariate analysis. Consequently, proportion of velar length to velar depth and velar ascent were associated with nasalance scores significantly.

1.运动分析的发展应用同源建模、FED和主成分分析利用三个红外相机和一个彩色相机生成耳朵运动范围图像，并将红外红色图案投射到脸上，然后对正常人进行嘴唇运动分析。在计算每个元音发声时表示嘴唇运动特征的主成分分数后，在二维上绘制主成分分数。判别分析显示，可以检测到发/u/ 和/i/ 时嘴唇运动之间的差异。因此，认为主成分分析对于唇部运动分析是有用的。 2．使用立体内窥镜生成距离图像使用直径为 5.4 毫米并包括两个望远镜的立体内窥镜开发了测量系统。初步研究揭示了立体内窥镜存在以下三个问题。 1）图像中心偏离。 2) 畸变变大，以致离开图像的中心。 3) 两幅图像的亮度存在差异。 Tsai算法应用于1）和2）的问题，LoG滤波应用于3）的问题。通过对立体内窥镜获得的图像进行校正和Birchfield算法，可以成功地生成距离图像。 3．使用腭咽头影图测量腭咽形状和鼻音评分之间的关系通过多变量分析研究了吹气和发声/a/时的腭咽头影测量值与从鼻音计获得的鼻音评分之间的关系。因此，软腭长度与软腭深度的比例以及软腭上升与鼻音分数显着相关。