A Testbed Combining Visual Perception Models for Geographic Gaze Contingent Displays

Abstract

We present a testbed featuring gaze-contingent displays (GCDs), in which we combined multiple models of the human visual system (HVS) to manage the visual level of detail. GCDs respond to the viewer's gaze in real-time, rendering a space-variant visualization. Our testbed is optimized for testing mathematical models of the human visual perception utilized in GCDs. Specifically, we combined models of contrast sensitivity, color perception and depth of field; and customized our implementation for geographic imagery. In this customization process, similarly to the geographic information systems (GIS), we georeference the input images, add vector layers on demand, and enable stereo viewing. After the implementation, we studied the computational and perceptual benefits of the studied perceptual models in terms of data reduction and user experience in geographic information science (GIScience) domain. Our computational validation experiments and the user study results indicate the HVS-based data reduction solutions are competitive, and encourage further research. We believe the research outcome and the testbed will be relevant in domains where visual interpretation of imagery is a part of professional life; such as in search and rescue, damage assessment in hazards, geographic image interpretation or urban planning.