A Study on Eye-Blink Detection-Based Communication System by Using K-Nearest Neighbors Classifier

There are 450 thousand patients in the world who is diagnosed Amyotrophic Lateral Sclerosis. These patients lose control of all the muscles in their body except the eye movements. The purpose of the studies in this research area is to design a system, which the patients can express the basic needs, works with high accuracy, which all patients can afford financially, with minimal discomfort to person. The aim of the developed system is to analyze the eye-blinks made with binary coding and convert them into sound. First, eye-blink signals coded with binary number sequence is obtained with NeuroSky MindWave Mobile device. These coded signals are given as an input to the designed system. In the eye-blink analysis section, the location and number of eye-blinks are determined. Then, the eye-blinks containing coded information in the input signal are classified using the K-Nearest Neighbors algorithm, and the class of the eye-blink is determined. Finally, the letters corresponding to the code of the eye-blink sequence are obtained and the resulting word is vocalized. With its low cost and high accuracy, this system causes minimal discomfort to the patient and has a simple structure.