Signal and Data Processing Journal A scientific journal officially licensed by the Commission for Scientific Publications of the (MSRT). Publisher: Research Ceter for Developmen of Technologies
Volume 23, Issue 1 (6-2026)
JSDP 2026, 23(1): 1-18 |
Back to browse issues page
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Soltanizadeh H, Mizban M, Zahraei M. A Hybrid Gaussian-Transformer Framework for Interpolation of Missing Data in Kinect Sensor. JSDP 2026; 23 (1) : 1
URL: http://jsdp.rcisp.ac.ir/article-1-1460-en.html
URL: http://jsdp.rcisp.ac.ir/article-1-1460-en.html
Semnan University
Abstract: (11 Views)
Accurate gait analysis plays a crucial role in diagnosing neuromuscular disorders and improving rehabilitation programs. However, motion capture sensors like Kinect often suffer from missing data due to technical limitations such as environmental noise, physical obstructions, and rapid movements. Traditional interpolation methods struggle with the complexity of multidimensional and nonlinear motion data, leading to suboptimal performance.
This study introduces an innovative hybrid interpolation framework that combines Gaussian Process (GP) for uncertainty modeling with a Transformer model for capturing complex spatiotemporal dependencies. The outputs of these models are optimized using Particle Swarm Optimization (PSO) and fused through weighted integration to enhance accuracy.
The proposed method was evaluated on motion data from 53 participants, with missing points artificially simulated for assessment. Results indicate that the hybrid method achieved a mean error of 0.0026, demonstrating a significant improvement over conventional approaches such as inverse distance weighting and random forests.
Key innovations of this framework include the use of multi-head attention mechanisms in the Transformer model for long-range dependency modeling, advanced RBF kernels in GP for balancing smoothness and adaptability, and intelligent optimization techniques for noise reduction and stability enhancement.
While computational efficiency requires further refinement, the proposed method marks a substantial advancement in precise motion data analysis for medical, sports, and rehabilitation applications. This research paves the way for integrating deep learning with statistical methods in biometric data processing, facilitating the development of real-time intelligent motion analysis systems.
This study introduces an innovative hybrid interpolation framework that combines Gaussian Process (GP) for uncertainty modeling with a Transformer model for capturing complex spatiotemporal dependencies. The outputs of these models are optimized using Particle Swarm Optimization (PSO) and fused through weighted integration to enhance accuracy.
The proposed method was evaluated on motion data from 53 participants, with missing points artificially simulated for assessment. Results indicate that the hybrid method achieved a mean error of 0.0026, demonstrating a significant improvement over conventional approaches such as inverse distance weighting and random forests.
Key innovations of this framework include the use of multi-head attention mechanisms in the Transformer model for long-range dependency modeling, advanced RBF kernels in GP for balancing smoothness and adaptability, and intelligent optimization techniques for noise reduction and stability enhancement.
While computational efficiency requires further refinement, the proposed method marks a substantial advancement in precise motion data analysis for medical, sports, and rehabilitation applications. This research paves the way for integrating deep learning with statistical methods in biometric data processing, facilitating the development of real-time intelligent motion analysis systems.
Article number: 1
Keywords: Missing Data Interpolation, Gaussian Process, Transformer, Kinect Sensor, Gait Analysis, Motion Data Reconstruction, multi-head attention mechanisms
Type of Study: Research |
Subject:
Paper
Received: 2025/02/23 | Accepted: 2026/02/8 | Published: 2026/06/21 | ePublished: 2026/06/21
Received: 2025/02/23 | Accepted: 2026/02/8 | Published: 2026/06/21 | ePublished: 2026/06/21
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |