Enhanced combined eye gaze direction classification and head flexion detection system

In the past years, the detection and tracking of eye-gaze has been considered an active research field due to the convenience it adds to a variety of applications. In addition, head movement detection has received similar attention. Both of these technologies can be used as alternative methods for interfacing devices. Several approaches have been presented in the literature and used in the implementation of many algorithms for these technologies. An enhancement to an eye gaze direction classification system is presented in this paper. Viola-Jones face detector is used in the eye gaze direction classification system for face detection and to define the eye region. However, Viola-Jones face detector has a limitation when detecting rotated faces. In the proposed approach, the Circular Hough Transform is used for locating the eye iris. Low level features are then computed and support vector machine (SVM), which was found to achieve the highest accuracy in the system, is used to classify the eye gaze direction. The proposed enhancement suggests calculating the head flexion angle using flex sensors and a microcontroller. The calculated flexion angle can then be used to rotate the face back to 0° angle. This overcomes Viola-Jones limitation regarding detecting rotated faces. It also increases the system accuracy when used with rotated face images. In addition to increasing system effectiveness in classifying eye gaze direction, determining the head direction is useful to be used as a control method. This combination of eye gaze direction classification and head direction detection provides various control signals that can be used in many applications such as assistive technology.