Real-Time Foot Tracking and Gait Evaluation with Geometric Modeling

Area of Science:

• Biomechanics
• Computer Vision
• Rehabilitation Engineering

Background:

• Gait evaluation is vital for monitoring patient recovery and balance in rehabilitation.
• Vision-based systems offer portable and low-complexity alternatives for gait analysis.
• Existing systems may have limitations with overground gait assistive devices due to restricted views.

Purpose of the Study:

• To develop and validate a novel vision-based foot tracking algorithm for gait analysis.
• To cater the algorithm for overground gait assistive devices with limited user view.
• To provide a contactless, portable, and cost-effective gait evaluation method.

Main Methods:

• The algorithm models the user's foot and shank using geometric principles.
• Cost optimization aligns these models to point cloud data of the lower limbs.
• The system tracks foot poses to compute spatial-temporal gait parameters.

Main Results:

• The algorithm demonstrated fitting rotational and translational errors below 20 degrees and 33 mm at 0.4 m/s.
• Gait detection achieved an F1 score exceeding 96.8%.
• Step length, step width, and cycle time errors were approximately 35 mm and under 38 ms, respectively.

Conclusions:

• The proposed algorithm offers a fast, contactless, and portable method for gait evaluation.
• It effectively computes spatial-temporal gait parameters without requiring specialized footwear.
• This technology supports gait rehabilitation and assistive device applications.