Font Size: a A A

Research On Reconstruction And Recognition Method Of Writing Trajecttory Based On Inertial Measurement

Posted on:2020-07-22Degree:MasterType:Thesis
Country:ChinaCandidate:C C CuiFull Text:PDF
GTID:2428330590973281Subject:Control engineering
Abstract/Summary:PDF Full Text Request
In this paper,a hardware device based on inertial measurement and its related trajectory reconstruction algorithm are proposed for motion trajectory reconstruction and handwritten digit recognition algorithm.The inertial measurement hardware device consists of a three-axis accelerometer,two gyroscopes,a microcontroller,and a wireless transmission module.The inertial device collects the motion data during the writing motion through the motion sensor,transmits the motion data to the computer through the wireless transmission module,and preprocesses the collected motion data,and then reconstructs the writing trajectory by using the quaternion-based trajectory reconstruction algorithm..In order to reduce the systematic error and drift error of the motion data,the mechanical arm is used to calibrate the gyro's zero offset and scale factor to reduce the system error.The Kalman filtering algorithm is used to process the drift data of the gyroscope to reduce the drift error of the motion data.The advantages of inertial measurement based hardware devices and their trajectory reconstruction algorithms are: the device is portable and can be free of any external reference device or writing range limitation;the quaternion-based trajectory reconstruction algorithm can effectively reduce the direction and integral error,Accurate reconstruction of the motion trajectory.Our experimental results have validated the effectiveness of inertial devices and their equipment.After reconstructing the writing trajectory,the convolutional neural network algorithm is used to train and recognize the written digital trajectory,and the written digital trajectory is successfully recognized.
Keywords/Search Tags:Inertial measurement, Quaternion, Inertial sensor, Convolutional neural network, Trajectory reconstruction algorithm
PDF Full Text Request
Related items