Eye Tracking in Geoscience: Data Registration and Visualization

Bo Hu¹, Tommy P. Keane¹, John A. Tarduno², Brandon B. May¹, Nathan D. Cahill^1,3, Robert A. Jacobs⁴, and Jeff B. Pelz^1
1Center for Imaging Science, Rochester Institute of Technology, Rochester, NY, USA
²Department of Earth and Environmental Sciences, University of Rochester, Rochester, NY, USA
³Center for Applied and Computational Mathematics, Rochester Institute of Technology, Rochester, NY, USA
⁴Department of Brain and Cognitive Sciences, University of Rochester, Rochester, NY, USA

Abstract

Geologists routinely infer geodynamic history by viewing landscapes, but the expertise for correct interpretation requires years to develop. Eye movement data, such as gaze timings, area of interests (AOI), and order of interpretation, provide insight in understanding viewing strategies developed in experts that may be valuable in the education of novices. Toward this goal, we are studying the eye movements of novice and expert geologists during an annual class fieldtrip at numerous sites in central California and Nevada using advanced portable mobile eye- trackers and high-resolution imaging systems.

In this article, we describe algorithms for mapping videos from multiple eye-trackers onto a single high-resolution panoramic image of a scene. The algorithms are solely image-based, comparing the outward-pointing videos captured from the portable eye-trackers with the high-resolution panoramic image using standard ideas from feature estimation and multiple view geometry, without requiring any extrinsic method to relate the positions of the eye-trackers.

The high-resolution scene images, from which the panoramas are created, can be also used in creating immersive "virtual field trips". We have found that simple rectilinear projections of the panoramas do not provide the correct perspective and sense of scale, and therefore have limited value in research and education. We describe a software framework that displays the images with correct viewing parameters, mimicking faithfully the view of the original scenes. In addition, the software tools allow intuitive annotation of AOIs by geologists, facilitating analyses of eye-tracking data.

Eye movement data provide valuable information in understanding viewing patterns of novice and experts. However, data collection in natural geological settings is demanding. Here we discuss the analysis and visualization challenges. We present an approach involving algorithms and software tools to register eye movement data in a common reference frame, and to effectively visualize and annotate high-resolution imagery. Application to our growing expert/novice data set will be discussed.

AAPG Search and Discovery Article #120140© 2014 AAPG Hedberg Conference 3D Structural Geologic Interpretation: Earth, Mind and Machine, June 23-27, 2013, Reno, Nevada