Ocular Motion from Facial Image Sequence to Detect Human Alertness

In the area of vehicular safety much attention has been drawn towards detection of the level of alertness in drivers (operators) to prevent road casualties. In this respect, author has developed image processing methods that estimate Percentage Closure of eye over time and saccadic ratio to determine the present state of alertness and to indicate the onset of drowsiness through day and night. The performance of the algorithms is validated on a set of videos containing facial images by correlation analysis with blood bio-markers and subjective questionnaire. To realize the procedure, author has designed a few experiments to induce drowsiness in human vehicle drivers in both day and night driving condition for production of multiple cues of alertness including the ocular indices.