Hand Gesture Recognition And Applied Research Based On 802.11a Long Training Sequence

With the development of computer technology and artificial intelligence,gesture recognition technology has become one of the hot spots in the research.Gesture recognition technology can be used to achieve efficient intelligent human-computer interaction system,which facilitates the exchange between people and machines.The characteristics of long training sequence of WiFi signals and pattern recognition method are used in this themes and a novel gesture recognition method based on WiFi signals is proposed.The method analyzes 802.11a long training sequence,extracts relevant features and uses pattern recognition methods to train model and recognize gestures.In addition,this paper has designed a practical application scenario interactive system.The main research findings of this paper are as follows:This paper analyzes the characteristics of 802.11a frame and long training sequence and proposes a feature extraction method based on 802.11a long training sequence.The long training sequence of 802.11a frame can be used to estimate the channel frequency responses and frequency offset.Different gestures could cause certain impacts on wireless signals.The effects will correspondingly change the frequency offset.Besides.The gestures can also cause certain impacts on the transmission channel of wireless signals and a corresponding change in the channel parameters.Therefore,this paper uses feature extraction methods to extract offset estimation and channel estimation parameters by using WiFi signals,and then merge the features.We use the autocorrelation algorithm to calculate the offset estimation by using long training sequence.Then we use channel estimation algorithm to calculate the channel estimation parameters.The simulation shows that the frequency offset and channel estimation parameters features have good performance which can significantly identify different gestures.After extracting the feature,this paper uses SVM classification algorithm to analyze the simulation results.The classification model has good classification performance,the test results show that a set of eight static gestures can be identified and the average recognition rate is 95%.The optimized average recognition rate is 97%.Based on above research,a human-computer interaction system is designed in this paper.The system can collect data and train model,analyzes and processes the received WiFi signals to extract the frequency offset and channel estimation parameters.Then the system uses the trained classification models to classify the gestures and show the results on the GUI.