A PCA-based technique for compensating the effect of sensor position changes in motion data

Failure to place markers/sensors accurately when motion capture the human body is probably the single greatest contributor to measurement variability in rehabilitation applications. In this paper, we study the effect of inadvertent changes in the position of sensors while evaluating similarities between similar actions in motion capturing. We then use a Principal Component Analysis (PCA) based technique followed by additional mechanisms to compensate for the effect of sensor position changes within the motion data. The collected data from our designed experiments show differences between similar actions with different marker placement. By applying signal processing, we show how these variations can be compensated for and so provide more accurate motion data.