Spontaneous Gaze-Based Application Using 2D Correlation and Velocity Threshold Identification

Demand for touchless technology has grown with the surge of Covid-19 pandemic. Spontaneous gaze-based application is one of several promising technologies for touchless interaction. Despite of this potential, little attention has been paid to the performance of traditional eye movements classification methods on improving accuracy of gaze-based object selection. To handle this research gap, we proposed a novel workflow of spontaneous gaze-based object selection using 2D Correlation as similarity measure and Velocity Threshold Identification (I-VT) as a method for eye movements classification. We compared our method with Pearson Product-Moment Correlation (PPMC) as similarity measure and Dispersion Threshold Identification (I-DT) for eye movements classification. Our experimental results showed that the proposed method yielded object selection accuracy up to 95:62%+/-3:48%. In future, our proposed method can be implemented in the development of touchless interactive technologies that adhere to the World Health Organization guidelines, especially during the Covid-19 pandemic.