Abstract
In stabilometry, the sway of the human body in an upright posture is studied by monitoring the displacement of its centre of pressure in the lateral (x) and anterio-posterior (y) directions. The area covered by this trace has been defined as that of an ellipse fitted to the data. Conventionally, its angle of inclination is found through linear regression (LR) on the data in the x - y plane. In the present paper, principal component analysis (PCA) is proposed as providing a more suitable basis for the estimation of angle and area. Results of simulations and stabilometric tests confirm large differences between area and angle estimates obtained by regression of x over y, and y over x, with PCA generally agreeing with either one or the other of the LRs. The PCA technique is therefore recommended as an improved basis for measuring area and inclination of stabilograms, or similar data sets.
Keywords: body balance, stabilometry, signal processing, principal component analysis