A robust cursor activity control with Iris gesture and blink detection technique

Abstract

Using computers for people with very limited or no arm movements is a very tough task and most of the time impossible. In this paper, we are proposing an iris based cursor control system developed speci cally for physically challenged people. Eye gaze localization has been a very popular research eld for several years. Tobi Eyes, Gaze Pointer Face Detection with viola-jones are the advanced works in this eld but both have certain limitations such as the infrared ray issues and accuracy deviation because of unstable frames. In our work, we have developed a model to overcome these barriers of the state of the art systems. Along with the software model we are also proposing a hardware model in order to improve the accuracy of previously mentioned di culties. In our system, we used Hough circle transform for localizing the eye gaze. The developed system is tested by multiple users and we achieved around 89.8% accuracy. This model will help the disabled peoples to get themselves out from their limited world. Furthermore, the functionality of using the on-screen keyboard will help them to interact very easily. This system will ensure maximum user comfort with minimum setup requirements.