Reducing Error Rate for Eye-Tracking System by Applying SVM

Abstract

Electrooculography (EOG) is widely considered the most effective signal-processing technique for identifying distinct eye movements. The EOG signal was used to extract functionality to provide dependable assistance to visually impaired patients. In EOG studies, the extraction of new features is an adequate and reasonable phenomenon. The EOG system is less expensive than any other signal-processing system. Still, it has significant drawbacks, such as a high error rate. In our study, we measured the Euclidean distance error. We found that it is 3.95 cm, which is significantly less than the standard error rate. The main objective of our study is to investigate an EOG analysis with the least possible error rate. EOG is substantially less expensive than other eye-tracking systems, and the proposed method can be used to provide a consistent user experience for visually impaired patients at a low cost with a minimum error rate. Moreover, this method can be applied in drone controllers, mouse controllers, and wheelchair controllers.