Constant and Variable Time Delays of Synchrony of Facial Expressions in Dyadic Conversations.
Constant and Variable Time Delays of Synchrony of Facial Expressions in Dyadic Conversations.
复制标题
DOI：
10.1080/00273171.2016.1265436
复制
发表时间：
2017-01
影响因子：
3.8
通讯作者：
Meyer MJ
中科院分区：
文献类型：
--
作者：
Meyer MJ
面部模型（组合外观模型，Cootes，Edwards & Taylor，2001）被用来研究在先前记录的2到4分钟的二元对话中几对人的面部表情的同步性。在每一个对话中，一个人被选为发言人。演讲者描述了他们感到快乐、悲伤、惊讶或失望的记忆，每个面部模型由99个点组成，代表面部的主要特征。构建来自模型的点矩阵，使得这些矩阵的列表示模型中的X和Y点坐标，并且行表示原始跟踪视频中的每个帧。为了保证面部表情的分析，这些矩阵的非刚性头部运动分量提取通过采取奇异值分解的模型。然后，基于同一对话中的人的模型的重建矩阵被水平连接以形成配对矩阵。除了静态面部表情之外，为了测试面部运动的同步性，还计算了点的无符号均方根（RMS）速度。RMS速度矩阵的构造类似于点矩阵。
Face models (combined appearance models, Cootes, Edwards & Taylor, 2001) were used to study synchrony in the facial expressions of several pairs of people in previously recorded 2-to 4-minute dyadic conversations. In each conversation, one person was chosen to be the speaker. Speakers described a memory in which they felt happiness, sadness, surprise, or disappointment.Each face model consisted of 99 points that represent the major features of the face. Point matrices from the models were constructed such that the columns of these matrices represented the X and Y point coordinates in the model, and the rows represented each frame in the original tracked video. To ensure that facial expressions were analyzed, the nonrigid head movement components of these matrices were extracted by taking the singular value decomposition of the models. Reconstructed matrices from the models based on people within the same conversation were then horizontally concatenated to form paired matrices. To test synchrony of the movement of the face in addition to static facial expressions, the unsigned root mean square (RMS) velocities of the points were also calculated. The RMS velocity matrices were constructed similarly to the point matrices.
影响因子：
7
作者：
Boker, SM;Xu, MQ;King, K
通讯作者：
King, K
DOI：
10.1109/34.927467
发表时间：
2001-06-01
影响因子：
23.6
作者：
Cootes, TF;Edwards, GJ;Taylor, CJ
通讯作者：
Taylor, CJ