E T V I S I O N M A N U A L
24
4.
Repeat steps 2 and 3 for the other three points. Note that the eye pupils must be recognized
for all of the points. The corneal reflections should be recognized on the bottom 2 calibration
target points, but often may not be recognized on the top two points, and this is OK.
5.
Check for satisfactory performance over the desired visual field region as previously
described. If satisfactory, click the ”Calibrate” icon to close the calibration procedure
(“Calibrate” icon turns black) and proceed to the next section.
The procedure described above should almost always result in good measurement accuracy with
in the visual field enclosed by the boundary connecting the outer 4 calibration points. Accuracy
will often be quite good significantly beyond this range, but error may sometimes increase
significantly as the subject looks progressively farther away from this region. In this case,
accuracy can usually be improved, before closing the calibration procedure, by adding an
additional 4 points approximately in the areas shown by the red circles, below. Simply repeat the
procedure described above for these points before clicking the Calibrate icon to close the
calibration procedure.
Red circles show suggested areas for calibration target points 6-9.
With default settings, the maximum number of calibration points is 9. (This maximum value can
be changed as described in Appendix B). Additional left clicks on the Scene window, beyond 9
points, will result in the point closest to the click being replaced by the new click. Once the
Calibrate procedure is closed (black “Calibrate” icon), the current calibration cannot be modified.
If the calibration process is re-opened, by clicking the “Calibrate” icon, the current calibration
result is deleted and the process begins again “from scratch”.
After each calibration point is entered (left mouse click), a “confidence” value for that pont
appears in the scene window title bar. Proper interpretation of the “confidence” value is discussed
in Appendix B.
