Network Teaching Contexts Learners Face Detection And Facial Feature Extraction

The research of this paper is supported by the key science program of Beijing Education Committee and by the key program of the natural science foundation of Beijing.(No. KZ200810028016).Personalized Network Education System,which has the function of mood-alternation,has become an important direction of the development of modern age network education.As the external performance of the students' study-mood,study-expression shows the subjective study status of the students.The function of study-expression recognition is used in the network education system,which is the foundation of the Personalized Network Education System's establishment.The main work of this paper is to extract the expression feature parameters for the analysis of the students' study-expression and the establishment of study-mood model,according to the image recognition,which is taken from the study scene.The most main feature parameters are near-far degree parameter,the distance of eyelids,mouth corner radian,etc.The detail work of this research is listed as below:1.Face detection.This is the foundation of the expression features extraction and even the whole model building.After researching on the existing algorithms of face detection,based on the improved algorithm of illumination compensation,a method based on the movement detection,skin-color model and neural network confirmation is proposed.Then we discuss the movement forecast and how to judge the state of study.The data of facial outline can be extracted after the detection.2.Distance of eyelids extraction.After researching on the existing algorithms of eyes location,we use the Geometry characteristic to delimit the eyes' possible area,then compute the complex degree to acquire eyes candidates.Finally,the two eyes can be located by the Integral projection.The distance of eyelids can be extracted by the searching algorithm of color characteristic.Finally we discuss whether the eyes are open or closed,and when the eyes are closed,is it shows that the student is blinking or sleeping?3.Mouth corner radian extraction.Considering the purpose of this research,a research plan is formulated,which is to extract the left mouth corner point,right mouth comer point and the center point of the lower lip.The possible area is delimited according to the mouth geometric distribution in the face,then formulate the research algorithm by the feature color characteristic. Finally,the mouth corner radian can be computed by the three points' coordinate.4.Based on those methods above,we develop an expression features extraction system by Visual C#.NET.There are such functions below:the camera takes photos in front of the computer on time,then the system can makes the judgment whether the student is on the scene; if the student is studying,the system extracts his expression feature data.The further work is to research the study-expression recognition deeply,based on the improvement of the algorithm's precision and the system's real-time function.