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

Three-Dimensional Analysis of Orofacial Movements during Pronunciation of a Sentence

句子发音过程中口面部运动的三维分析

基本信息

批准号：
04454490
负责人：
YAMAGATA Kensuke
金额：
$ 2.56万
依托单位：
Showa University
依托单位国家：
日本
项目类别：
Grant-in-Aid for General Scientific Research (B)
财政年份：
1992
资助国家：
日本
起止时间：
1992 至 1994
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-04454490/
关键词：
Speech Motion analysis High-speed video Facial movement Mandibular movement Retro-reflective marker Auto tracking Sound spectrograph

项目摘要

We developed a cordless light-emitting, retroreflective target marker that produces a clear shining point when illuminated by a cold spotlight. This enabled the recording and autotracking of the multiple facial targets on an image processor. The facial targets and the voice of each subject were simultaneously recorded on the same video tape by two high-speed TV cameras, each focused on a different aspect of the face. The image and its voice components were respectively fed into an Image Data processor (ID-8000) and a DSP Sona-Graph. Each timeframe during which each consonant was pronounced was identified and numbered on the the basis of its waveform and/or sound spectrogram. The movement of each facial target during each time-frame was tracked automatically by the Image Data processor and the 3-dimensional coordinated data was transferred to a computer.The trajectory of each monitored target during pronunciation of each tested consonant was computed by a 3-D analyzer (Movias 3D) with reference to the following parameters : the total length from start to finish of the trajectory (TL) ; the distance between the start and finish of the trajectory (SL) ; the ratio of the TL to the SL (T/S) ; the volume of the rectangular solid encompassing the entire trajectory (cubical range) ; and the mean of the 3-dimensional angles caused by differences in the direction of the preceding and following trajectories at each measured time point (TH).The advantages of this method are that it gives an actual sequence of the real images in motion, not just a symbolic or mathematical representation of movement.

我们开发了一种无绳发光，反光目标标记，产生一个明确的亮点时，由冷聚光灯照明。这使得能够在图像处理器上记录和自动跟踪多个面部目标。每个受试者的面部目标和声音都被两台高速电视摄像机同时记录在同一盘录像带上，每台摄像机都聚焦在面部的不同方面。将图像和其语音分量分别馈送到图像数据处理器（ID-8000）和DSP Sona-Graph。每个辅音发音的时间段都根据其波形和/或声谱图进行识别和编号。图像数据处理器自动跟踪每个面部目标在每个时间帧内的运动，并将三维坐标数据传输到计算机，由三维分析器计算每个测试辅音发音过程中每个监测目标的轨迹（Movias 3D），参考以下参数：轨迹从开始到结束的总长度（TL）;轨迹起点和终点之间的距离（SL）; TL与SL的比值（T/S）;包含整个轨迹的长方体的体积（立方范围）;以及在每个测量时间点（TH）由在前和在后轨迹的方向上的差异引起的三维角度的平均值该方法的优点在于，它给出了运动中的真实的图像的实际序列，而不仅仅是运动的符号或数学表示。