学习情景中表情特征数据的提取研究

Research on facial features extraction in the study scene

论文讨论了在学习情景中如何提取表情特征数据的方法。提取的数据包括眼帘间距和嘴角弧度,分别用于关注度与愉悦度的研究。在人脸图像上,先使用几何特征法划定出人眼的大概区域,再通过复杂度的计算与积分投影定位出双眼,以色调为依据制定搜索算法获得眼帘间距。在定位出双眼的基础上,进行嘴角与下唇中心点的定位,由三个特征点的坐标值计算出嘴角的弧度。实验证明该方法对不同的人脸图片有较好的适应性,较准确地提取出了所需的数据。

This paper discussed the algorithms for Facial Features Extraction in the study scene. One extracted data is the distance of eyelids which is used in the attention degree, the other data is the mouth corner radian which is used in the joyful degree. In the face image, geometry characteristic method is used to find the possible eyes areas first. Then locate the two eyes by computing the complex degree and the center degree, and extract the distance of eyelids by research algorithm. Based on eyes location, two mouth corners and the center point of the lower lip can be located. Through the three feature points, the radian can be computed. The experiment showed that the algorithms are suitable for different face images in the study scene and extracted the data accurately.