Biometric system for measuring gait and fall characteristics captured on video

Area of Science:

• Biomechanics and Human Movement Analysis
• Computer Vision and Image Processing

Background:

• Quantifying gait and fall characteristics from video is crucial for understanding human movement and injury prevention.
• Traditional methods often struggle with real-world complexities and accurate 3D reconstruction from 2D video.

Purpose of the Study:

• To develop and validate a robust methodology for quantifying gait and fall characteristics using video analysis.
• To introduce a new metric, the stepping/impact angle, for analyzing fall dynamics.

Main Methods:

• Applying principles of projective geometry, specifically homography, to transform 2D video frames into real-world measurements.
• Utilizing an optimization algorithm to enhance the accuracy of on-screen point selection for the homographic transformation.
• Validating spatial gait parameters against the GAITRite system, achieving a limit of agreement of -2.12 cm to 2.03 cm for step length.

Main Results:

• The developed method accurately quantifies spatial gait parameters from video footage.
• The analysis of real fall events revealed pre-imbalance gait characteristics and stepping during the fall.
• The stepping/impact angle effectively quantifies the influence of stepping on fall direction.

Conclusions:

• The proposed video analysis methodology provides a reliable tool for quantifying gait and fall biomechanics.
• This approach can be valuable for clinical assessments, injury prevention research, and understanding fall mechanisms.