The development of processing- and assessment tools for event detection in eye movement data

The development of processing and assessment tools for event detection in eye movement data

With the rapidly growing number of people using eye trackers, the need to address and solve fundamental problems related to the analysis of eye-tracking data have become increasingly important. Fixation, saccades, and smooth pursuit are prototypical eye-movement events that comprise fundamental units in eye-tracking research, since they provide direct links to a variety of cognitive processes. Previous research has identified a number of serious flaws in current algorithms that are used to detect these events. A particular challenge is to detect such events in mobile eye trackers, when participants are free to move their heads.

The purpose of this project is to devise improved definitions, methods, and algorithms for event detection. Specifically, we will investigate techniques for noise reduction and artifact removal in raw eye-tracking data, design filters that allow accurate calculation of eye-movement velocity and acceleration, and develop new event detection algorithms. In particular, we will pioneer smooth pursuit algorithms, which make it possible to identify periods when the eye follows a moving target; both in data collected with stationary and mobile eye trackers. The current lack of such algorithms has limited the possibility to perform experiments with moving stimuli in, e.g., research about human computer interaction and cognitive psychology.

Eye-, and head-movements are measured when participants perform a variety of tasks. Knowledge about the physiological limitations of eye movements, the technical properties of the eye trackers combined with advanced signal processing methods are used to develop improved event detection methods.

Research group

PI:Prof. Marianne Gullberg
Lund University Humanities Lab, Lund University
Linnea Larsson
Dept. of Biomedical Engineering, Lund University
Martin Stridh
Dept. of Biomedical Engineering, Lund University
Marcus Nyström
Lund University Humanities Lab, Lund University

Links and references

Larsson, L. (2016)
Event Detection in Eye-Tracking Data for Use in Applications with Dynamic Stimuli .
Doctoral dissertation, Lund University

Larsson, L., Nyström, M., Andersson, R., & Stridh, M. (2015)
Detection of fixations and smooth pursuit movements in high-speed eye-tracking data
Biomedical Signal Processing and Control, 18, 145-152

Larsson, L., Nystrom, M., & Stridh, M. (2013)
Detection of saccades and postsaccadic oscillations in the presence of smooth pursuit
Biomedical Engineering, IEEE Transactions on, 60(9), 2484-2493