Research And Implementation Of Gesture Recognition Algorithm Based On Inertial Sensor

With the rapid development of computer technology and the widespread popularity of smart devices,people have put forward higher requirements for the efficiency and the convenience of Human-Computer Interaction(HCI).Hand Gesture Recognition(HGR)technology based on inertial sensors only requires the user to hold the device and draw simple gestures to complete complex machine commands.This method does not distract the user's visual attention and is not affected by external environmental noise and light intensity.Hence,it is more suitable for smart wearable devices.In order to realize the gesture recognition function on smart student card products,MPU6050 is selected as the inertial sensor,which can output 3-axis acceleration data and 3-axis gyroscope data;n RF52832 is selected as the development board,which has a programmable flash memory in the 512 KB system and 64 KB system RAM.In order to improve the accuracy of gesture recognition accuracy,the inertial sensor is calibrated beforehand,the weighted recursive average filtering is used for data denoising,also the differential absolute average method is used for gesture endpoint detection(ED),and the zero padding method is used to pad the data,and the quaternion complementary filtering method is used to calculate the device's position,and the equivalent rotation vector method is used to update the quaternion.The restricted coulomb energy neural network algorithm is applied to gesture recognition.The traditional RCE algorithm has the problem that it is difficult to control the number of neurons and excessive overlap between neurons.Combined with the characteristics of the time correlation of the sensor gesture sequence,the RCE algorithm has been improved in two aspects: the element reliable value estimation method and the distance method between neurons.In order to verify the performance of the algorithm,a 3D Digital Dataset containing 10,000 samples has been made,and the 5-fold cross-validation method is used for experimental verification,and the average gesture recognition accuracy has been improved as high as 98.68%.