Polynomial approximations of gait for human motion analysis and visualization

Human motion can be digitally encoded following video motion capture analysis. This process can be computationally intensive or require human intervention when a large number of frames are considered. Data reduction methodologies using cubic interpolation polynomials were investigated. This was done to reduce the number of frames which must be considered by motion analysis software and to reduce bandwidth requirements for collaborative visualization across the Internet. Cubic spline control points were initially selected using the start, left swing, left stance, right swing, right stance, and end position of the gait cycle. An adaptive algorithm is presented to add control points as required to bring the error, calculated as a percentage of joint range, to within a maximum threshold value. It was determined that for subjects with normal gait, six to eight control points were required to generate curves with the appropriate tolerances. Editing polynomial curves by moving control points to visualize the net effect on gait of changes in joint range following surgical procedures are also discussed