Gesture Recognition Glove System Based On Graphene Flexible Sensors

With the rapid development of modern computer technology,the application of artificial intelligence becomes more and more extensive.The interaction between human and computer has become an indispensable part of people's daily life.Therefore,more and more researchers are beginning to explore the field of human-computer interaction,and have also made many new discoveries.The research on gesture recognition is an important development direction in the field of human-computer interaction.In particular,the gesture recognition based on data gloves is a hot topic of research.Although the vision-based gesture recognition has high precision,it has more variability in complex environments,and the equipment is also more expensive.So this text mainly studies the gesture recognition based on the data glove,under the premise of low cost,this kind of recognition is not easy to be influenced by the environment,and can guarantee the recognition rate.On the basis of summarizing the status quo of gesture recognition research,this paper analyzes two kinds of gesture recognition methods,studies several algorithms,and focuses on the gesture recognition of self-made data gloves.The main tasks are as follows:Firstly,for the problem of low collection efficiency,this paper studies the characteristics of the graphene flexible sensor and uses it as the data acquisition sensor.This data glove is equipped with a self-made data acquisition module,this module has a STM32 series microcontroller as its core component.Secondly,based on this data glove,the sensor bending data is collected by the acquisition program and noise processing circuit to the computer.Then the data is calibrated and stored as a gesture recognition database.The experimental results show that the graphene sensor and self-made data gloves can improve the data collection efficiency and ensure the training data volume.Thirdly,this paper uses three gesture recognition algorithms,template matching method?BP neural network and two-stage method.The template matching algorithm is written in the C language environment.The mean value of each column of data in the database is used as a gesture template.The Euclidean distance is calculated by matching the collected new data with the template data,and the specific gesture value of the data to be identified is obtained.The neural network algorithm is written in the MATLAB environment.First,the data is normalized,then the input and output neurons are established,and finally the data to be recognized is input to the net to obtain the gesture recognition rate.When we find that the above two algorithms have their own defects,we propose a two-stage method that combines the steps of the two algorithms.Firstly,we use template matching to obtain high-precision data that can be identified,and then input this part of the data into the neural network to obtain a higher recognition rate.Finally,based on the above research basis,this paper prepares a host computer interface program based on Java,collects real-time gesture data,recognitions gesture values and displays the experimental results.Based on the data gloves,the gesture recognition rate of the three algorithms is 93.3%?96.7% and 98%.It can be judged that the two-stage method is more suitable for the identification of this topic.