Research On Learners' Emotion Recognition Method In Teaching Environment

Due to the large number of classes in traditional classroom teaching,a teacher cannot effectively and effectively perceive the emotional state of multiple students in the learning process.Learners' cognitive barriers to new knowledge may result in negative emotional states,which may lead to a decrease in learning efficiency.Teachers' timely adjustment measures based on students' learning emotions,and enhance classroom-emotional communication between teachers and students,will help improve classroom learning.Teaching efficiency.The use of information technology to help teachers obtain timely information on students' emotional status in the classroom is an important research content in the current education field.Facial expressions are rich in emotional information and can reflect people's psychological state and mood changes.At present,most researchers study based on static facial expression recognition.However,facial expressions of learners are a dynamic process in the process of teaching.Expression recognition technology based on static images will not be applied to the actual teaching environment.In order to solve the above problems,this dissertation researches on dynamic serial expression recognition technology,and realizes the detection of the current emotional state of learners through the recognition of dynamic facial expressions.In the research process,this paper firstly uses the emotional ring model as a theoretical basis to construct a learner's emotion model to describe the mapping relationship between the six basic expressions and emotional states.Secondly,taking into account that manually extracting the peak frame of a sequence of expression images is not conducive to practical application,the slow feature analysis algorithm(SFA)is used to obtain the peak expression frame automatically.Again,in order to obtain the key feature information of the facial expression change,this paper calculates the data of various geometric shapes between facial feature points in the sequence image and maps the data.According to the characteristics of the data change in the map,the feature of the change is more obvious,that is,the key characteristic that contributes greatly to the facial expression,mainly including the data of the four types of angles,slopes,Euclidean distances,and polygons.Then,Gabor transform is used to extract facial texture features,and the fusion features are used as facial expression feature data.Finally,in order to solve the problem of effective fusion of heterogeneous feature data,shallow learning is difficult to improve the accuracy of expression recognition,and the problem of deep network learning architecture applied to the classification of small sample data sets,this paper proposes an expression classification method based on deep multiple kernel learning.For six kinds of basic expressions,the classification method proposed in this paper is based on the The Extended Cohn-Kanade Dataset(CK+)dynamic sequence expression library experiments,and the obtained face facial expression recognition accuracy rate can reach 94.4%,indicating that the method proposed in this paper is for facial expressions The improvement of the recognition rate has a certain role,which lays a solid technical foundation for the later application of teaching and promoting the intelligentization of the teaching environment.