Research On Technology Of Gesture Recognition Based On Smart Wearable Controller

With the development of wearable electronic devices,human-machine interactions based on gesture recognition has become a research hotspot.In order to miniaturize the size of wearable devices and optimize gesture recognition algorithm,a gesture recognition system based on wearable controller is proposed.By using the inertial sensor to collect the gesture data,the recognition result is displayed on the upper computer by the gesture recognition algorithm,which can be worn on the finger and achieve the purpose of random input of human-computer interaction.The controller is composed of main control module,attitude module,communication module and power-supply module.The main control module uses Atmeg328 embedded microprocessor.Motion data acquired by attitude module of MEMS inertial sensors(Micro-Electro-Mechanical System)accelerometer and gyro inertial sensor is received by the embedded microprocessor,which makes date coordinate transformation and kalman filtering and exports gestures data to PC via WIFI protocol.The power module supplies power to the whole system and charge the lithium battery according to the selected mode.The system is running on the Processing software based on java,The acceleration and angular velocity data of gesture are intercepted by difference score,six basic features is extracted,according to the characteristics of each gesture to determine the threshold value of characteristic quantity,the gestures was classified by a four 4 layer decision tree classifier combined with characteristic quantity extracted and its threshold.Gesture recognition results is displayed by humancomputer interface in real time..The whole process does not need testers to conduct pre-training without any template matching and building,to reduce the system difference and calculation.Results based on 20 experiments,when the sampling frequency of the wearable gesture recognition controller is 50 Hz,achieves an overall recognition accuracy of 97.17%,gesture recognition accuracy is from 95% to 100%,this design of gesture recognition system based on wearable controller improves the accuracy of gesture recognition algorithm.