A novel 3D vibration monitoring method using a monocular camera and phase difference Gradient-based algorithm

• Published in: Measurement : journal of the International Measurement Confederation Vol. 248; p. 116909
• Main Authors: Yang, Xiasen, Yan, Peng, Liu, Xiao, Lu, Wenbo, Zhang, Xiangyu, Chen, Ming, Wang, Gaohui
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15.05.2025
• Subjects: Aperture problem; Monocular camera; Phase-based optical flow; Three-dimensional vibration; Vibration measurement; Vibration measurement; Phase-based optical flow; Aperture problem; Monocular camera; Three-dimensional vibration
• ISSN: 0263-2241
• DOI: 10.1016/j.measurement.2025.116909

•A novel 3D vibration monitoring method using a monocular camera and PDGB algorithm is proposed.•Out-of-plane vibrations are reconstructed using phase difference gradients, eliminating in-plane motion errors.•PDGB algorithm applicability is analyzed with respect to filter orientation and pixel changes.•The accuracy and robustness of the proposed method are validated through simulation and laboratory experiments. To address the increasing demand for large-scale dynamic vibration monitoring, machine vision-based methods are gaining prominence, while accurately estimating three-dimensional (3D) vibrations remains challenging. This paper introduces a novel 3D vibration monitoring method utilizing a monocular vision system. The phase difference gradient-based algorithm is proposed to characterize out-of-plane motion and correct for in-plane motion estimation errors. A simulation experiment was conducted using synthesized 3D vibration video to validate the method, assessing the effectiveness of filter orientations and pixel changes. Additionally, real-world validation was performed using a concrete vibration table to demonstrate the accuracy and robustness of the proposed method. Results indicate that the proposed method can precisely measure 3D vibrations with a resolution of 0.01 cm under the test conditions. The robustness of the proposed method is further confirmed across varying light intensities, active pixel selections, and noise levels, highlighting its potential for dynamic response monitoring applications.