Research On Interactive Method Of Gesture Motion Track Based On Motion Capture

The common input method of computer information input is to write on a fixed plane.By collecting the data information written in the plane,and then transmitting the data to other devices for processing,such an input mode belongs to two-dimensional spatial information transmission.The purpose of this paper is to study the input of information in three-dimensional space.The MEMS inertial sensor is used to write numbers and characters in three-dimensional space and reproduce the writing track.In this paper,a motion track detection system based on MEMS acceleration sensor is designed.The advantages of this system are small size and low cost,which is in line with the development trend of handwriting trace detection system at home and abroad.The system consists of two parts: data acquisition and processing module and upper computer digital track recognition module.The data acquisition and processing module is composed of arduino101 and IMU.The upper computer digital track recognition and implementation is completed by MATLAB.The data acquisition and processing module is mainly responsible for sensor data acquisition,filtering,attitude fitting,attitude correction,quaternion coordinate conversion,integral capture trajectory and data integral compensation.Finally,the captured carrier trajectory data is sent to the upper computer through the serial port.The upper computer system is mainly responsible for converting the trajectory received by the serial port into a picture and training the neural network,and recognizing the picture by gray-scale processing,binary processing and neural network classification algorithm after image reduction.The system chooses nine axis sensor mpu9250 to detect the acceleration information and angular velocity information generated when the sensor moves in the air.Its principle is that the acceleration data integration is velocity,and the velocity integration is displacement.In spatial handwriting,the direction will change.At this time,quaternion can be calculated by Euler angle.Quaternion can convert the three-axis acceleration under the carrier coordinate system to the three-axis acceleration under the geographic coordinate system,and then coordinate conversion can fix the acceleration data in the same coordinate system.These displacement points can be connected to get the track information of aerial handwriting.In the process of measurement,because of the influence of the error,the problem of motion track display error is caused.The Kalman filter is designed and implemented to reduce the random error.Part of the data in the open source handwritten digital data set is used as the training set,and the self-made handwritten data is used as the test set to train and verify the accuracy of neural network recognition.The experimental results show that the method of inertial trajectory capture and neural network recognition can effectively recognize hand gestures.150 samples of handwritten numbers are recognized,and the recognition accuracy is 100%.The accuracy of trajectory recognition is 60% for capturing the number of inertial trajectory,which achieves the expected effect.