Empirical Evaluation and Optimization of Fisheye Views in Dynamic Visual Search and Selection Tasks

Fisheye views, providing local details and global context, are widely used to mitigate unintended activations in gaze interaction. By magnifying targets near the gaze position, they can improve target discriminability and reduce demands for precise gaze positioning. However, their performance in dynamic interfaces remains constrained by instability and inappropriate parameter settings. This study investigated the effects of fisheye view type, dwell threshold, and target motion velocity on gaze interaction performance in a dynamic search and selection task with 30 participants. Accuracy rate and response time were measured. Results showed that standard fisheye magnification improved selection reliability without significantly enhancing efficiency. A dwell threshold of 0.5 s balanced accuracy and response time. Magnetic fisheye produced the most stable interaction under low-speed, moderate-threshold conditions, whereas bubble fisheye enabled rapid access but was more prone to unintended activation at short dwell thresholds. These findings inform fisheye optimization for dynamic gaze interfaces.