Research On Gesture Recognition System Based On WiFi Device

Contactless gesture recognition is an emerging interactive technique in the ubiquitous and mobile computing.This new technology combines the linguistics with the wireless signals to analyze,judge and integrate human gestures by the usage of intelligent algorithms.The existing contactless gesture recognition studies can achieve the gesture recognition with machine learning technologies.But in practice,some objective factors,such as the user's position,the non-line of sight condition,can seriously affect the performance of these gesture recognition systems.In this paper,by analyzing the existing research on gesture recognition methods at home and abroad,the traditional gesture recognition algorithms require specialized equipment,installation troubles,and the use of machine learning methods to place higher demands on the human body.Combined with physical layer channel state information Based on the characteristics of Fresnel zone model,a gesture recognition system is constructed,and a complex gesture recognition is realized by using hidden Markov model.The main work and innovations of this article include the following aspects:This paper studies the effect of different human gestures on the signal in the Fresnel zone under indoor WiFi signals.According to the change of the signal in the Fresnel zone,this paper proposes a gesture recognition system based on the physical layer information of the WiFi signal.It is a fine-grained and location-independent intelligent and robust contactless gesture recognition system.The gesture recognition system proposed in this paper can realize common gesture recognition based on Fresnel region model and decision tree classifier algorithm.Based on this system,a complex gesture is considered as a combination of simple gestures.Hidden Markov models are referenced in the gesture recognition module to build a complex gesture recognition system based on hidden Markov models to achieve complex and continuous gestures.We design and implement the prototype system based on commodity WiFi cards.We take comprehensive experimental evaluation in different indoor environments.Experimental results show that the simple gesture recognition system achieves strong robustness and high accuracy of 91% on average.The recognition rate of complex gesture recognition system based on Markov models is as high as 86%.The research in this article helps to theoretically reveal the performance potential of CSI and Fresnel zone technologies in wireless environment sensing applications,and explore the feasibility of a simple gesture recognition system for humans based on Fresnel zone models.The feasibility of human complex gesture recognition system based on Hidden Markov model was explored.