Research And Implementation Of Hand Gesture Recognition Terminal Based On MEMS Sensors

Human-computer interaction(HCI)technology has constantly changed people's lives.Hand gesture,as an important way of human-computer interaction,has become the focus of research.Currently,hand gesture is recognized mainly via tactile,visual and motion sensors.However,the way of tactile sensor is limited by the size of the touch screen,and the research based on visual sensor is dependent on external conditions and limited by the camera range.Compared with two other methods,the gesture recognition based on motion sensor has many advantages,such as low cost,less affected by surroundings and convenient to be implemented.In this thesis,a hand gesture recognition terminal without excessive sensors and peripheral devices has been designed using MEMS sensors to recognize hand gestures online,which is small,easy to wear,low-cost,and unlimited scope of recognition.The main works are listed as following:(1)Hardware design of the terminal is implemented,including analysis of hardware framework,determination of the working flow and the selection requirements of components,lectotype of the hardware modules and design of the related circuits.(2)Three types of hand gestures are defined,whose recognition algorithms,including collection of gesture data,interception and preprocessing of effective gestures,and establishment of decision-making methods,has been designed.Firstly,gesture data is collected using MEMS sensors.Secondly,a three-threshold forward-backward search method is proposed to intercept effective gesture data.Then,the data exception filtering and smoothing filtering are applied to the gesture data.Lastly,the two-level classifier is designed to make decision.In the first level,the decision tree algorithm is used to identify the categories of the three types of gestures.In the second level,specific gestures of certain category are identified.(3)The software framework is designed by the strategy of layering.The framework has four layers,which are the drive layer,the interfacing layer,the system abstract layer,and the application layer.The drive layer makes hardware work normally.The interfacing layer serves as the interface of drive layer.The system abstract layer is designed by state pattern where every state of the terminal can switch to each other.The application layer is designed by state pattern and strategy pattern where hand gesture algorithm runs.Additionally,experiments were carried out to verify and analyze from three aspects: recognition rate,real-time and energy consumption.Test results show that the terminal has an average recognition rate of 89.3%,system delay of less than 100 ms,endurance time of 140 hours in standby state and 48 hours in working state.In the practical test,the terminal is used to control home equipment effectively.