Skip to Main Content
In this paper, a method to create an application which is competent of replacing the traditional input device (mouse) by using human facial features is proposed. Distinctively, using real time videos of the user's face extracted from the video sequence obtained using an off-the-shelf web-camera. It can be applied as an optional input source for those who cannot use their hands due to disabilities or patients who cannot use their hands. In the proposed technique, a method that combines both feature-based and image-based approach is used. The fundamental approach for detection is fast extraction of face candidates using Six-Segmented Rectangular (SSR) filter and then pass them to Support Vector Machine for face verification. In face tracking, the patterns of between-the-eyes are tracked with update template matching. A window that has the feature's template size is scanned over the Region of Interest (ROI) and then calculates the Sum of Squared Difference between a frame that has the feature's template and the current frame. Experiments show that 90% of the system behaves satisfactory for a web-camera at frame rate of 15 fps with the image resolution of 320 times 240 frame size. The system consumes little amount of CPU resources allowing other processors to run smoothly.