Optimum object selection methods for spontaneous gaze-based interaction with linear and circular trajectories

Abstract

The demand for touchless technologies has increased since the outbreak of Covid-19. Spontaneous gaze-based interaction is a promising technique because it offers a natural approach for object selection using eye movements. There are two types of object trajectory in spontaneous gaze-based interaction: linear and circular trajectories. However, previous studies did not consider the optimum object selection method by taking into account the characteristics of each trajectory. In addition, previous studies suffered from high missed detection that caused spontaneous gaze-based applications to be less accurate and responsive. To address these scientific gaps, we conducted a novel study to propose optimum methods for object selection in linear and circular trajectories. The experimental results suggested that each trajectory requires a different combination of events detection and object selection techniques to achieve optimum performance. We recommend 2D Correlation as the object selection technique. In addition, we also recommend the Velocity and Movement Pattern Identification (I-VMP) and the Hidden Markov Model (HMM) as events detection techniques in linear and circular trajectories, respectively. The proposed approaches solved the missed detection problem and significantly increased the accuracy of gaze-based object selection to 95.60% and 99.73% on linear and circular trajectories, respectively. In the future, our discoveries are promising for the development of accurate, responsive, and seamless gaze-based touchless applications.