| With the rapid development of science and technology,the emergence of various intelligent devices has made people’s lives more intelligent.Human-Computer Interaction has also become an integral part of people’s daily lives.As an important means of HCI,human gesture recognition technology has strong theoretical research and practical application value,which has become a research hotspot in various fields.With the development of wireless technology,the coverage range of WiFi signals is getting wider.Therefore,using ordinary WiFi signals to sense human gesture become a new way of intelligent sensing.In recent years,many human gesture recognition systems based on WiFi signals have appeared at home and abroad.With the deepening of research and the maturity of technology,the recognition accuracy of each system has reached a high level,and there have been certain improvements in efficiency and stability.However,there are still problems of limited recognition positions and poor system portability.The method based on Weighting Principal Component Analysis(WPCA)proposed in this paper effectively solves the problem of human gesture recognition in the area(position-free).The main work can be summarized as follows:1)The Intel 5300 network card is used to build the platform for collecting Channel State Information(CSI)data,which works in Monitor mode.The network card working in the Monitor model can control the parameters,such as the number and speed of data packets,the transmission channel and bandwidth.At the same time,the equipment is more stable and the anti-interference ability is enhanced.2)To make the system achieve better denoising effect,a joint denoising scheme based on low-pass filtering and Principal Component Analysis(PCA)method is proposed in this paper.To extract more representative features,this paper uses First order Difference(FoD)and Discrete Wavelet Transform(DWT)methods for feature extraction.Among them,the FoD can remove some multipath interference,and the DWT performs time-frequency analysis on the waveform to extract more fine-grained features.3)By researching and analyzing the impact of PCA method on human gesture recognition at different positions in the area,a recognition system based on WPCA method was designed.According to this effect of different principal components have different human gesture recognition capabilities for different positions,the method assigns different weights to each principal component,which improves the system’s recognition accuracy effectively.In order to further improve the recognition ability of the system,the entire recognition area is divided into 9 areas in this paper,and training a model in each area.Firstly,based on the area location algorithm proposed in this paper,the user’s position in the area is located using the leading gestures,and then the area model is used for human gesture recognition.The algorithm successfully improves the system recognition accuracy to 95%.Through repeated comparison experiments,it is proved that the recognition effect of the proposed algorithm in this paper is better than the existing ensemble algorithms,and the robustness of the system is verified by small sample data and Signal Noise Ratio(SNR)curve analysis. |
