Monday, July 24, 2017

Day 13—7/24/17

In the morning I worked on finding some saccades on the angular velocity graphs and understanding what they looked like.  Saccades occur when the eye moves from one fixation point to another at a high rate.  For example, when you notice a small movement in your peripheral vision, your eyes jerk to that point of movement that you had sensed.  To find what a saccade looked like on the angular velocity graphs, I first found a few by watching the eye tracking video of the data collected and noting the times of the saccades that I noticed.  Then I went to the angular velocity graphs and found the intervals on which I had noted saccades.
After this, I learnt about how to "clean" the data and make the graphs smoother.  One of the reasons why the graphs were not very smooth and had many outliers was due to the fact that when we blink, the eye tracker cannot track the eye movements, so it predicts the gaze vectors of each eye.  This prediction sometimes results in the eye tracking data stating that the z-coordinate of the gaze vector was negative, and this implies that the wearer's eye was angled towards the inside of the head.  To correct for this, we can interpolate for the gaze vectors with invalid z-values.

Here is an image of the angular velocity graph (the saccades are circled in green and the outlier is circled in red)

Here is a short video of some of the eye tracking data (the orange circle represents the point at which the eyes are directed at)

No comments:

Post a Comment