| Title |
Ground-Truth Data Modeling for Gait Analysis based on Kinect V2 Sensor Data |
| DOI |
https://doi.org/10.5573/ieie.2022.59.8.99 |
| Keywords |
Gait analysis; Kinect sensor; Multivariate multiple regression analysis; Ground-Truth data; Modeling |
| Abstract |
Gait analysis is being used in various fields including human identification, locomotion assessment, rehabilitation, and diagnosis for individuals with cerebral palsy, spinal cord injury, and stroke, and degenerative brain diseases. Ground-Truth data in gait analysis are obtained from Vicon or Optotrak systems using optical motion capture methods. However, those systems have limitations in that they require markers, installation space, and significant cost. To compensate for these disadvantages, recent research in gait analysis investigates as an alternative the application of body point position data in 3D space from a system consisting of multiple Microsoft Kinect sensors. The previous studies focused on the linear correlation, reproducibility, and concordance between the Ground-Truth data and the data from their proposed systems in evaluation. In this study, a method of modeling and predicting Ground-Truth data based on data from a multiple Kinect system using a multivariate multiple regression analysis model is considered, and the correlation is analyzed by computing the coefficient of determination and the intraclass correlation coefficient between the fit values and the actual Ground-Truth data. |