Validation study of a Kinect based body imaging system.

Area of Science:

• Biomedical Engineering
• Human Factors Engineering

Background:

• Three-dimensional (3D) body scanners are increasingly explored as replacements for traditional anthropometric data collection methods.
• Ensuring the reliability and precision of 3D scanner data is crucial for widespread adoption in diverse applications.
• Existing research has primarily focused on white light and laser-based scanners.

Purpose of the Study:

• To comparatively analyze anthropometric data acquired via a Kinect body imaging system against traditional manual methods.
• To compare Kinect system results with findings from previous studies on various 3D body scanners.
• To assess the impact of participant movement and stance variation on data accuracy.

Main Methods:

• Calculation of Mean Absolute Difference (MAD) for anthropometric measurements.
• Comparison of MAD values against the maximum allowable error stipulated in ISO 20685.
• Evaluation using a physical mannequin to isolate the effects of body movement and stance variation.

Main Results:

• Few body measurements met the restricted allowable error threshold (ISO 20685).
• Measurements from a physical mannequin yielded results closer to the ISO 20685 limit compared to human participants.
• Despite improvements with the mannequin, results still exceeded the allowable error margin.

Conclusions:

• The primary source of discrepancy between 3D scanning and manual methods is the data acquisition time.
• Involuntary body sway in human participants significantly impacts accuracy, especially with longer scan durations.
• Kinect system accuracy is compromised by participant movement, necessitating further research for real-world applications.