Research And Implementation Of The Micro-expression Recognition Method

With the advancement of global information,educational information has become an important way of promoting the modernization of education in China.The college plays an important role in promoting the development of educational information,it is very urgent for colleges to apply information technology to various components of talents training,to improve the quality of talent training.The micro-expression is the expression and disguise of the individual's internal emotions,which is difficult to forge and suppress.Therefore,it can be used as the evaluation clue of multiple educational links.Psychological health education,as an important part of college education,has increasingly important influence on improving the quality of talent training.To provide important basis for the following education,many colleges use online Chinese college students' psychological health appraisement system at present,which is one of core work for the information construction in the field of psychological health appraisement.However,this appraisement collects the answers of college students only at the scale aspect,which leads to the low reliability of appraisement due to individual subjective cognition.Thus,this paper takes micro-expression as the research object,and studies the detection and recognition of micro-expression in video.And in scene of college students'psychological health appraisement,the application of micro-expression recognition in college students' psychological health appraisement is studied.The main research works of this paper are as follows:(1)Research on ROIs(Regions of Interesting)extraction for micro-expression recognition.First,four micro-expression ROIs are divided according to the characteristics of the muscle strain movement of different micro-expressions.Then,the training dataset of ROIs extraction model is constructed.What's more,the effects of weight initialization,learning rate and training iterations on the training effect of the model are explored through a large number of contrast experiments with the control variables.Finally,an ideal ROIs extraction model is trained.(2)Research on micro-expression recognition based on ROIs.Firstly,all video fragments that might contain micro-expression among video sequences are screen out with ROIs extraction model.And based on the common spatial features of micro-expressions and macro-expression,a micro-expression recognition model through transfer learning is constructed.And after amounts of experiments,weights and parameters of this model are modified.Finally,by comparing the performance of this algorithm with the existing excellent micro-expression recognition algorithms in the same database,the progressiveness of this algorithm is verified.(3)Development of the new college students' psychological health appraisement prototype system.According to the characteristics of college students' psychological health appraisement and based on micro-expression recognition and the online psychological health appraisement technology,a new college students' psychological health appraisement prototype system is designed and developed,which provides a new information technique and tool for the appraisement with high validity.It is of practical significance and application value for perfecting the educational information system.