Driver’s Gaze Zone Estimation Method Based On Monocular Camera

According to the study, most traffic accidents are caused by driver’s inattention. Eye tracking technique has a rapid development over the past ten years, and it plays an important role especially in driver’s attention and fatigue detection. Driver’s gaze zone estimation becomes the hot spot and key point to solve traffic accident nowadays.Currently, most eye tracking methods have good effects in the laboratory environment. But when driving, there is a series of complex scenes like strong light and large rotation of the head, most methods of gaze zone perform not well. And nowadays most methods of driver’s gaze zone estimation use head posture to estimate gaze zone only.To solve the problem of driver’s gaze zone estimation, we proposes a method that combines head pose and the gaze angle. The method include five patch, (1)we adopt the monocular camera to get driver’s head image, then we use face detection method to get ROI.(2)On the ROI we use SDM to extract human face feature points. (3)And then we use the optimization POSIT method to detect the driver’s head pose, our method is more accurate and stability. (4)As for the eye area, we use the method of hexagon qualified to find the ROI and the pupil center according to the gradient information, then the gaze angle is estimated from 3D eye model, last, we use linear SVM to detect glasses to get confidence parameters of view. (5)The gaze angle is calculated by combining line of sight and head pose.The paper complete a real-time systems observing the driver’s gaze zone esti mation and analyses the driver’s driving behavior. According to the system results show that the proposed method can be in the driver’s complex environment has a higher accuracy. The method proposed in this paper is able to detect vehicles in real time, which makes a good foundation for driver assistance system.