Markerless identification of key events in gait cycle using image flow

Area of Science:

• Biomechanics
• Computer Vision
• Human Motion Analysis

Background:

• Gait analysis is crucial for understanding human locomotion.
• Traditional methods often require specialized equipment like markers or calibrated cameras.
• There is a need for accessible and markerless gait analysis techniques.

Purpose of the Study:

• To develop and validate an image-flow-based method for automatic gait analysis using a single video camera.
• To identify key events within the gait cycle (e.g., toe-off, heel-strike) without markers.
• To assess the potential of this method for unencumbered human locomotion capture.

Main Methods:

• Utilized image-flow techniques to compute body segment motion and velocities from video.
• Implemented automatic segmentation to isolate subjects from the background.
• Analyzed head excursion and shank motion to identify gait cycle events.
• Validated results against the Optotrak 3D motion capture system.

Main Results:

• The proposed method successfully computed motion and velocities of body segments.
• Key gait events were accurately identified and extracted from video images.
• Results showed good agreement when compared to the Optotrak 3D motion capture system.
• The approach demonstrated markerless and unencumbered human locomotion capture capabilities.

Conclusions:

• Image-flow-based gait analysis offers a viable, markerless alternative for motion capture.
• This technology has potential applications in home and community-based gait monitoring.
• It can serve as an assessment tool for gait symmetry and establishing normal gait patterns, particularly for the elderly and disabled.