期刊名称：International Journal of Computer Science and Network Security
印刷版ISSN：1738-7906
出版年度：2017
卷号：17
期号：4
页码：182-191
出版社：International Journal of Computer Science and Network Security
摘要：Presently, eye tracking is a one of the most important domain of research of computer vision dealing with the detection of eyes and associated parts, especially the corners, centres etc. to identify the point-of-interest of any person. This methodology has huge potential for application for the patients who have lost their communication abilities to communicate with the outside world using eye tracking. A large number of eye trackers are available commercially, but they failed to become popular due to higher cost and complexity. As a result, researchers are engaged around the world to develop low-cost eye tracking systems using most available components in the market. But, none of these systems are evaluated against different difficult and critical situations which they may come across. Such evaluation of the basic underlying methods may save time, effort and cost towards successful technological development. The present work evaluates the performance of an eye tracking methodology to be used to develop a rehabilitation system for stroke patients under different difficult and critical situations, such as different illuminations during day and night, low light during night, users with and without spectacles, variable distances between the camera and users etc. The optimum range of operation, minimum value and type of illumination, orientation of the light sources has been established through experimentation.
关键词：Face detection Eye tracking Haar-like features Hough circle Stroke Rehabilitation
