Design Of Recognition System And Study Of Quality Evaluation For Upper Limb Movement Based On Acceleration Infomation

As the development of sensor technology and pattern recognition technology, Human-Computer Interaction is now being developed from the traditional way which people adapt to computer to the the way which computer adapt to human. Although this new interactive way hasn't achieve the "people-centered"interaction of real sense, in Virtual Reality it has already cut a striking figure, especially in the game market it has won a great commercial success. The sensor applied in the game based on the new interactive way can capture one or several of EMG ,audio signal,video signal and acceleration signal, and the game based on the acceleration signal is the most successful in the game market of no doubt.However, all of the games based on the acceleration signal seriously affect the player's interest becauce of some disfigurement of them , such as in the "Wii Sports" game developed by a company in Japan players simply move the game handle but at the same time shown strong swing in the game content ; as well Lele Jian game and Jing Dongfang game,et al, can not truly identify the player's action because of some technical limitation. To solve these problems, this paper put forward a new program of acceleration signal acquisition based on the human motion model, and use this signal to achieve the identification of three conventional kinds of action from the game of table tennis by some model, the action consists of forehand drive, backhand drive and kill shot. This paper also attempts to study the means which is used for appraising the quality of upper limb movements. The main work of this paper is as follows:1,Analyse the model of human motion, put forward a new program to obtain the information of upper limb in play. The program uses two three-dimensional acceleration sensors, respectively placed in a fixed position of the forearm and rear arm , then in virtue of acceleration information acquisition system it can effectively obtain the information of the upper limb acceleration in the movement.2,Aimed at three kinds of table tennis action this paper establishes a support vector machine model of the ability of common generalization. Major step is described as follow: preprocess the acceleration signal to extract the effective action signal, by dint of normalization receive 6-channel acceleration signal of the same length, use wavelet transform to achieve feature extraction, and then use the ant colony algorithm for feature selection , at last establish support vector machine model based on feature vectors selected by ant colony algorithm.3,Use principal component analysis on the study of the quality evaluation of upper limb movement, this paper extracts the two main elements affecting the quality movement in this system. Extract 39 physical indicators from 6-channel acceleration signals, by use of standardization of these indicators, depending on the value and variance of these factors extract two principal components, by analysing the principal component receive two indicators which are explosive force and endurance ,these two indicators are the two main components which can evaluate the quality of upper limb movement in this system.