베이지안 네트워크에서 Kalman필터 알고리즘을 이용한 실시간 시선식별에 관한 연구

Abstract

Eye-gaze is an input mode which has the potential of an efficient computer interface. Eye-gaze has been the focus of research in this area. Non-intrusive eye-gaze tracking that allows slight head movement is addressed in this paper. A small 2D mark is employed as a reference to compensate for this movement. The iris center has been chosen for purposes of measuring eye movement. The gaze point is estimated after acquiring the eye movement data. Preliminary experimental results are given through a screen pointing application. Humans easily recognize where another person is looking at and often use this information for interspeaker coordination. We present a method based on three neural networks of th local linear map type which enables a computer to identify the head orientation of a user by learning from examples. One network is used for color segmentation, a second for localization of the face, and the third for the final recognition of the head orientation. The system works at a frame rate of one image per second on a common workstation. We analyze the accuracy achieved at different processing steps and discuss the usability of the approach in the context of a visual human-machine interface. The movement of user's eyes can provide a conventient, natural and high-bandwidth source of input. By tracking the direction of gaze of the user, the bandwidth of communcaiton from the user to the computer can be increased by using the information about what the user is looking at, and even designing objects specially intended for the user to look at. A variety of eye-gaze tracking techniques have been reported in the literature. A short list includes Electro-Oculography, Limbus, Pupil and Eyelid Tracking Contact. Lens Method, Corneal and Pupil Reflection Relationship, Purkinje Image Tracking, Artificial Neural Networks and Head Movement Measurement. Computer vision is intrinsically Non-intrusive, and does not require any overly expensive equipment. Non obtrusive sensing technology - such as video cameras and microphones - has received special attention in this regard. This paper draws on computer vision and image processing techniques for measuring eye-gaze.