Sufficent lighting.
this image was with two flashlights, centered off of the sclera (so that we would not get glared
spots on the eye): note how the pupil stands out wonderfully.
Here's the logic:
- In the multiple squares experiments, we showed that we could track the movement of the eye even as the eye
focused on squares as small as 1/256th of the screen. however, due to poor lighting, we couldn't find the
pupil.
- The above image shows that if we shine visible lights onto the eye, we can get the pupil clearly. However,
visible light directly into the eye is very painful, so we could not replicate the 256 squre experiment.
- The work of others shows that the quickcams can see in IR, and that IR leds from remote controls
can illuminate in IR the same way that visible flashlights illumnate in visible. Theoretically, we would
repeat the same experiment in IR, but we don't have the resources to spend $200 on an quickcam to perform the experiments.
We therefore design a system where we would illumante the eye with IR in the same fashion as we ilumated in visible, and
would create final placements based on what we discovered through those final experiments. Unfortunately,
we can't conduct those final experiments, but we can lay them out.
One big step toward making this work is knowing the IR spectral characteristics of
the quickcam. Here, courtesy of connetix, they are:
Camera Placement
the following three images show the results we got when we placed the camera in various locations:
|
|
|
|
|
The camera mounted under the eye.
|
The camera mounted flush with the forhead.
|
The camera mounted over the eye two inches away from the forhead.
|
These experiments lead us to conclude we should put the camera under the forhead. Here, the yellow
ball substitutes in part for the size of the camera casing.
Of course, the camera isn't going to be that large: since we can dispense with the ball of the casing,
the remaining part of hte camera is about the size of an AA battery, so the real imapct of the camera looks like
this:
Other future experiments
We will need some way of maintaining the ambient light level at the eye. We would do this by
1. Measuring the optimum light level near the eye when we get a great shot.
2. Mounting a photodetector near the eye, on the apparatus, to measure the ambient light and raise
the led lights accordingly.