Iris Localization and Tracking for Eye Gaze Directed Light Beam In Automobiles
Front road visibility for drivers during night time is provided by headlamps of fixed coverage area. The Problem arises at the time of overtaking, when the light beam cannot reach the far left or far right of the vehicle due to fixed coverage area, hence constraining the driver visibility. In this paper, we propose an adaptive forward lighting in automobiles based on the line of sight of the driver. Hence the headlamp can adapt to the driver`s view angle and direct the beam to the direction of the view. Both the eye centers (Iris) of the driver is continuously tracked to determine the view angle of the driver. The estimation of eye center is done by using image gradients which contains only dot products that corresponds to a point where most of the gradient vectors intersect and hence arrive the eye center. The eye center estimation algorithm is invariant to scale, rotation, pose and illumination. To increase the accuracy of the estimating the location of the iris, the nose is taken as a fixed reference point. The algorithm is implemented in BeagleBone Black processor and the feedback from the eye gaze is given to the headlamps by triggering the I/O pins of the processor.