Short communicationAccuracy map of an optical motion capture system with 42 or 21 cameras in a large measurement volume
Introduction
Optical motion capture (OMC) is used in a variety of fields, including biomechanics (Bell-Jenje et al., 2016, Le and Marras, 2016). The accuracy of various optical motion capture systems has been extensively evaluated in the past (Carse et al., 2013, Ehara et al., 1995, Ehara et al., 1997, Richards, 1999, Thewlis et al., 2013). However, these evaluations are typically performed in small volumes, from 0.005 to 15 m3 (Eichelberger et al., 2016, Windolf et al., 2008). Research on occupational ergonomics interventions, sports biomechanics, and rehab biomechanics often involve activities that span a larger volume, and thus require a more expansive motion capture system. Windolf and Eichelberger found that accuracy varies by location within the capture volume. In relatively large capture volumes (>100 m3), it is expected that accuracy has the potential to vary considerably by location, and a single measurement of accuracy at or near the center of the space is not adequate. No studies could be found that examined how accuracy changed approaching the edges of the capture space. Additionally, most previous studies have compared inter-marker distance to a more precisely known length of the same object. However, by only considering error in the measured distance (1D) between two markers, this technique fails to capture off-axis (3D) errors.
Thus, the aims of the current study were to establish the 3D accuracy of an OMC system for tracking individual markers within a large capture volume that is currently being used for biomechanical research, and to determine what portion of the capture area has acceptable accuracy for full-body biomechanics applications, such as gait analysis and occupational ergonomics.
Section snippets
Materials and methods
The OMC system used in this study was composed of 42 OptiTrack Prime 41 cameras and operated using OptiTrack Motive 1.10.1 Final software (NaturalPoint, Corvallis, Oregon, USA). Each calibration of the OMC system was performed by hand using an OptiTrack CWM-250 calibration wand with a length of 250.018 ± 0.002 mm. A panorama of the motion capture area can be seen in Fig. 1.
A ThorLabs LTS300 (Newton, New Jersey, USA) linear motion stage was used to evaluate the accuracy of the OMC data. The stage
Results
Maps of mean RMSE for each number of cameras are shown in Fig. 3. The majority of the capture area had an interpolated RMSE <200 μm (∼97% for 42 cameras; ∼91% for 21 cameras). Every point had a mean RMSE less than 1 mm, and the only locations that that exceeded 200 μm of error were those in the extreme corners of the capture area.
As expected, there is generally less error toward the center of the capture area. Errors then increase nonlinearly approaching the edges of the capture area, as revealed
Discussion
The results of this study confirm that OMC can be used to measure passive marker locations with accuracy better than 200 μm in the vast majority (97%) of even relatively large capture volumes (>100 m3), given optimal marker visibility conditions. Even when using only half (21) of the cameras, 91% of the capture area was found to achieve the same accuracy.
One objective of this study was to understand the magnitude of error contributed by the OMC system compared to other known sources of error when
Conclusion
The OMC system demonstrated submillimeter mean accuracy at every location in the capture volume, and error was found to be less than 200 μm in 97% of the capture volume (using all 42 cameras). Only very near the edges of the capture volume did error exceed 200 μm. The errors of the OMC system were found to be orders of magnitude smaller than other known sources of error associated with skin-mounted markers (marker placement errors and soft tissue artifact). Therefore, a large OMC system like the
Conflict of interest statement
OptiTrack (NaturalPoint, Corvallis, Oregon USA) provided equipment (including the linear motion stage and calibration wand) and technical assistance to optimize the system for this study.
Acknowledgments
None.
References (17)
- et al.
The association between loss of ankle dorsiflexion range of movement, and hip adduction and internal rotation during a step down test
Manual Ther.
(2016) - et al.
Affordable clinical gait analysis: an assessment of the marker tracking accuracy of a new low-cost optical 3D motion analysis system
Physiotherapy
(2013) - et al.
Human movement analysis using stereophotogrammetry – Part 4: Assessment of anatomical landmark misplacement and its effects on joint kinematics
Gait Posture
(2005) - et al.
Comparison of the performance of 3D camera systems. 2
Gait Posture
(1997) - et al.
Comparison of the performance of 3D camera systems
Gait Posture
(1995) - et al.
Analysis of accuracy in optical motion capture – a protocol for laboratory setup evaluation
J. Biomech.
(2016) - et al.
Comparative accuracy of radiostereometric and optical tracking systems
J. Biomech.
(2009) - et al.
Evaluating the low back biomechanics of three different office workstations: seated, standing, and perching
Appl. Ergon.
(2016)
Cited by (81)
Dynamic measurement method based on temporal-spatial geometry constraint and optimized motion estimation
2024, Measurement: Journal of the International Measurement ConfederationPractical whole-body elasto-geometric calibration of a humanoid robot: Application to the TALOS robot
2023, Robotics and Autonomous SystemsAn improved semi-synthetic approach for creating visual-inertial odometry datasets
2023, Graphical ModelsA hand motion capture method based on infrared thermography for measuring fine motor skills in biomedicine
2023, Artificial Intelligence in MedicineUsing an Optimal then Enhanced YOLO Model for Multi-Lingual Scene Text Detection Containing the Arabic Scripts
2024, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)