M.N. Pilipenko - Engineer, Bauman Moscow State Technical University, Moscow, Russia
E.Yu. Latysheva - Post-graduate Student, Bauman Moscow State Technical University, Moscow, Russia
I.N. Spiridonov - Dr.Sc. (Eng.), Professor, Head of Department, Bauman Moscow State Technical University, Moscow, Russia

Rehabilitation of people with disabilities is one of the trend of medicine at the moment. Special equipment for the care, orientation, training, orthopedic products, training and sports equipment develops to extend the functionality of the human possibilities. Pupil movement and facial expression can be used for screen point acquisition. With the advent of technology feature recognition of human motor activity, such as gestures, facial expressions, the movement of the pupils, the development of alternative methods of human-computer interaction became possible. For these purposes, there are a variety of hardware and software products: eye trackers, software (SW) for moving the cursor by the head. Eye trackers require expensive equipment or put on special glasses. Existing software, using an image from the DVR, reflects only the movement of the head and has a low accuracy of positioning the cursor. Proposed solution of automation of management human-s functions is to create a biotechnical system, which provides non-contact control the cursor by the direction of gaze. Webcam as a DVR permits to make the system less expensive. The aim of this work is to develop an algorithm for automatic pupil detection for automatic cursor targeting on the screen. There are various approaches for pupil detection in the video image. However, they have disadvantages like expensive equipments, contact technical means, time consuming to compute features and learning algorithms. In this work, the algorithm automatically determine the position of the pupil center using an adaptive binarization image with subsequent determination of the center of mass of the area corresponding to the position of the pupil. Average frequency of correct determination of the pupil center is 96.1%. The developed algorithm is optimal by the criterion of correct determination of the pupil center position with an accuracy of 5 pixels.