Using two overlapping or adjacent calibration frames to enlarge the 3D volume appears to provide accurate results. Errors for predicting points, were less than 0.5% of the calibrated volume along each axis if the predicted points were within the volume. Errors increased dramatically for predicting points outside the volume and thus care must be taken when using this technique to ensure that the enlarged volume fully encompasses the activity to be analyzed.
© Copyright 2000 All rights reserved.