A Custom Implementation of the Velocity Threshold Algorithm for Fixation Identification

Identifying fixations and saccades is an essential component of eye movement data analysis. There exist many algorithms that are employed to find fixations and saccades. In this paper we study the Velocity-Threshold Identification algorithm and implement it as an open source code for a specific dataset. A modified version of the algorithm was implemented and the number of fixations were compared with those reported by running a commercial software application. It was observed that the results are similar with an average difference of only 6 fixations per stimulus. The open source implementation also has a module to identify the centroids of fixation groups. This research aims to provide the open source community the ability to analyse an eye-tracking dataset without the use of any commercial software application.