Eye-Gaze and Vigilance Estimation |
Zakia Hammal
Corentin Massot
In the context of Human/Machine interaction, an efficient algorithm to iris segmentation and its application to automatic and non-intrusive gaze tracking and vigilance estimation is presented:
|
|
demo under construction
|
|
|
|
Abstract:
A luminance gradient technique is used to fit the irises from face images.
A robust preprocessing which mimics the human retina is used in such a
way that a robust system to luminance variations is obtained and contrast enhancement
is achieved. The validation of the proposed algorithm is experimentally
demonstrated by using three well-known test databases: the FERET database,
the Yale database and the Cohn-Kanade database. Experimental results confirm
the effectiveness and the robustness of the proposed approach to be applied
successfully in gaze direction and vigilance estimation.
Iris Segmentation:
The performances and the limits of the system and its ability to deal with different database are highlighted thanks of the analysis of a great number of results on two different databases : the Cohn-Kanade database and a database acquired at our laboratory (Hammal-Caplier database).
Figure.1 presents some examples of segmentation. For further details go to Hammal research
|
Figure.1 Results of iris segmentation.
|
2.Gaze Direction Estimation:
We have to
define the geometrical model and the projection function so as to establish the relationship between the position
of the iris center in an image and its projection on the screen.
|
Figure.2 System Overview and Geometrical Model.
|
In order to evaluate the
precision of our gaze direction estimation, an experimental setup, consisting in a grid
of 18 black points plotted on the screen (Figure 3) is used to estimate the user
gaze position during the fixation of these points. The subject was asked to sequencially fix the
black points and this experiment is carried out 10 times. The white
circles represent the estimated user mean fixation position for each fixed black point.
We obtain a mean precision of 0.8° which means that the fixations are accurately detected.
|
Figure.3 Results of the evaluation of the system precision and
exploration of a geographical map.
|
These results are very satisfactory in the context of human-computer interaction.
One of the potential application is the study of the strategy of a document exploration,
like in figure 4 which presents the analysis of a geographical map. The
system is able to detect the user attention during the exploration. Each point represents
the position of the gaze position on the map and its temporal apparition.
We then perform comparison experiments with a commercial system.
We present two kinds of comparison results:
a fixation map and a trajectory map.
In the fixation map, the user has to fix each
icon presented in the image in a free order. On Figure 5, each point represents one
fixation (barycenter of a region of fixation). The associated number indicates the order
in which the different icons have been looked at by the user.
In order to evaluate the performances of our detection system, we have made the
same experiments with the infra-red detector Eye-Link system (a commercial eye tracker
http://www.eyelinkinfo.com/) (Figure 5 right). The comparison of the detection obtained by both systems points
out a similar quality of our results.
|
Figure.5 Fixation map with our system (on the left) compared with a commercial system (on the right).
|
In the trajectory map,
we aim at rebuilding the ocular trajectory of a user. Here, he has to follow
the edges of a drawn house (figure 6).
|
Figure.6 Trajectory map with our system (on the left) compared with a commercial system (on the right).
|
2.Vigilance Estimation:
Associated to a blink detection system and a frequency analysis, our system is also able to detect the level of vigilance of the user in front of his computer screen.
For further details go to Hammal research
|
Figure.6 Vigilance estimation results on a sequence.
|
