Accuracy of our static posture classification system has been
evaluated with additional pressure maps collected from two
groups of subjects. First, an additional 200 pressure distribution
maps have been collected from 20 of the 30 subjects who
contributed to the static posture database (one sample per
posture per subject). These pressure maps are then labeled with
respect to their corresponding postures by the static posture
classification system. Fig. 7 shows the classification accuracy
in terms of percent-correct scores averaged over postures, as a
function of the number of eigenvectors