MEASUREMENT ERROR IN SIMULATED 20 GAIT DATA EXTRACTED BY A VIDEO-BASED MOTION MEASUREMENT SYSTEM

Abstract

An instrument reliability study was conducted in order to quantify the measurement error contained in 2D spatial data (linear and angular) commonly collected in gait and long jump studies (e.g. footfall position, toe-board-distance or limb orientation) by video-based motion measurement systems. Three experiments examined the effect of camera-to-object distance (5 to 10 m) and field of view (2.5 to 4.5 m) upon marker sets (stationary) representative of walking or running motion. The results show 7 to 10 m camera-to-object distances to be associated with the least error (mean absolute error = 0.59 cm) or about a third of the error found for the 5 m distance (mean absolute error = 1.5 cm). No systematic changes in measurement error were found across the differing camera fields of view with the mean absolute error falling below 0.66 cm and 0.6 .