Method Research Of Human Upper-limb Motion Capture And Recognition Based On MEMS Sensor

With the rapid development of Micro Electro Mechanical System(MEMS),the cost and volume of MEMS sensor devices reduce gradually,it has become preferred carrier equipments of the human motion capture and recognition research fields.Therefore,developing a appropriate method of data processing and application aim at the characteristic of MEMS sensor has become the primary problem to be solved for the field.In the human body movement,as the upper-limb movement has significant characteristics and most functional in the daily,so this paper mainly study around the upper-limb motion capture and recognition.Firstly,in order to be able to get closer to the human upper-limb movement actual situation of the data,in this paper,the data acquisition device fixed position and the data of the components are analyzed,determining the trend to pretreatment method and the wavelet threshold de-noising method.A wavelet threshold de-noising method has designed from the perspective of both qualitative and quantitative,and further analysis and verification has been made used by dynamic data.Secondly,this paper captures the movement of human upper-limb by capturing the attitude angle.At the completion of the study of the upper-limb attitude angle calculating method,through the capture of human upper-limb initial and dynamic attitude angle,so as to complete the research work of human upper-limb to capture.For this process,in this paper,visualization simulation model is built up based on VR technology is used to verify with real situation.Thirdly,the SVM is used to studying human upper-limb movement recognition.First,SVC for human motion recognition is designed based on SVM theory.Second,SVC is optimal designed by means of various optimization methods including the grid search algorithm and heuristic algorithm,for achieving better performance.At last,researching the division,extract and fusion methods of the movement data feature.Finally,on the basis of the research of human upper-limbs motion capture and recognition,formulating the corresponding experiment scheme.For the research of human upper-limbs capture,verification algorithm of motion capture first,and then design theupper-limb movements used in the experiment,at last,by comparing visual simulation model to validate the capture process.For the research of the upper-limbs recognition,designing a variety of classification model used in the experiment respectively,analyzing the results of experiment and draw conclusions.