Handwriting Trajectory Reconstruction Using Low-Cost IMU

• Published in: IEEE transactions on emerging topics in computational intelligence 2019-06, Vol.3 (3), p.261-270
• Main Authors: Pan, Tse-Yu, Kuo, Chih-Hsuan, Liu, Hou-Tim, Hu, Min-Chun
• Format: Article
• Language: English
• Subjects: Acceleration; Accelerometers; complementary filter; convolutional neural network; Discriminant analysis; Feature extraction; Handwriting recognition; handwritten letter recognition; Image reconstruction; Inertial measurement unit; Inertial platforms; inertial sensor; linear discriminant analysis; Low cost; Motion perception; Random noise; Reconstruction; Reconstruction algorithms; Sensitivity; Smart phones; Smartphones; Three axis; Trajectories; Trajectory; trajectory reconstruction
• ISSN: 2471-285X; 2471-285X
• DOI: 10.1109/TETCI.2018.2803777

In this paper, we propose a trajectory reconstruction method based on low-cost Inertial Measurement Unit (IMU) in smartphones. The IMU used in our work consists of a three-axis accelerometer and a three-axis gyroscope, which can record information of acceleration and rotation, respectively. Since intrinsic bias and random noise usually cause unreliable IMU signals, filtering methods are utilized to reduce high- or low-frequency noises of the signals. In addition, to more accurately detect whether the smartphone is moving or not, we extract multiple features from IMU signals and train a movement detection model based on linear discriminant analysis (LDA). Then, a "reset switch" mechanism is applied when the smartphone is detected as in a static state. The "reset switch" mechanism can effectively restrain the accumulated error of displacement calculation. Finally, the trajectory reconstruction results are applied to handwritten letter recognition for English alphabet, and the experimental results show that our proposed trajectory reconstruction method is reliable.