Unfortunately, it happens that some people lose ability to control their arms or legs because of accidences or diseases. Allot of people in World with physical impairments are in need of alternative input devices, which could help them to interact physical device without actually touching it. The communication system should use different actuating force such as speech, brain signals or the eye gaze. During last decade the interest in communicating technology, so called gaze tracking system, have been rapidly growing, because it consist of relatively inexpensive components and more important it is completely non-invasive. The gaze tracking systems detects the position of the user eye automatically using video camera and associates the control commands related with that position. Therefore, it is a novel and perspective way to communicate and control an application without a physical contact with real object. The presented work is dedicated to the research, design and realization of novel automatic gaze detection system, which is based on the solutions of image processing, pattern recognition and applications of the adaptive methods.