A warning system about drowsy status (fatigue or drowsiness) of the driver, helping to limit the traffic accidents caused by falling asleep behind the wheel by determining the status of the eyes combined with the head direction of driver. In this paper, we present a novel approach for determining the position of a human face and eyes in different directions of the face (turning left, turning right, tilt left, tilt right, looked-up, broke-down), using ASM model. Besides, we suggested an improved algorithm to detect and identify the status of the eyes to support the driver during driving. This algorithm consists of three main steps: (1) enhance the brightness of the image area contains two eyes using color spaces and median filter processing; (2) the image area contains two eye was extracted by a method of morphologies and image segmentation with dynamic threshold; (3) finally, the status of the eyes is determined by using SVM classifier based on HOG feature. Next, the direction of the driver's head is estimated by combining the 3D head model and algorithms posit to increase the effectiveness of the warning system. Our tested is performed with the data which was set by our (3778 photos of eyes of 9 people). Testing accuracy achieved is 97.86% and the average recognition time is 0.106s/image frame. Testing accuracy confirms the effectiveness and feasibility of the proposed system.