Non-rigid alignment pipeline applied to human gait signals acquired with optical motion capture systems and inertial sensors.

An accurate gait characterization is fundamental for diagnosis and treatment in both clinical and sportive fields. Although several devices allow such measurements, the performance comparison between the acquired signals may be a challenging task. A novel pipeline for the accurate non-rigid alignment of gait signals is proposed. In this paper, the measurements of Inertial Measurement Units (IMU) and Optical Motion Capture Systems (OMCAP) are aligned using a modified version of the Dynamic Time Warping (DTW) algorithm. The differences between the two acquisitions are evaluated using both global (RMSE, Correlation Coefficient (CC)) and local (Statistical Parametric Mapping (SPM)) metrics. The method is applied to a data-set obtained measuring the gait of ten healthy subjects walking on a treadmill at three different gait paces. Results show a global bias between the signal acquisition of 0.05°. Regarding the global metrics, a mean RMSE value of 2.65° (0.73°) and an average CC value of 0.99 (0.01) were obtained. The SPM profile shows, in each gait cycle phase, the percentage of cases when two curves are statistically identical and reaches an average of 48% (22%).