Real Time Visual Cues Extraction for Monitoring Driver Vigilance
ICVS '01 Proceedings of the Second International Workshop on Computer Vision Systems
Determining driver visual attention with one camera
IEEE Transactions on Intelligent Transportation Systems
Hi-index | 0.00 |
The authors analyzed facial video recordings and saccadic eye movements during 1-hour simulated driving in 10 subjects. Mean cross-correlation coefficient between the visually determined facial sleepiness and the proposed index of saccade (i.e. PV/D) for 9 subjects was -0.56 and the maximum coefficient of inverse cross-correlation was 0.83. Mean cross-correlation coefficient for 6 repetitive measurements for another subject was -0.72, and the maximum was 0.84. Variation in PV/D preceded that in facial sleepiness in 13 of 15 measurements and syncronized with it in other 2 measurements. From these results, we confirmed a fair potential of the PV/D to detect decline in vigilance levels earlier than facial sleepiness. We also revealed that narrow fluctuations throughout the measurement could lead to low inverse cross-correlation below 0.60 between the two indices. Therefore experimenter should pay attention to designing the experimental procedure to ensure broad fuctuations of the subject's vigilance levels in the measurement.