Project

Discipline

Geovisualization displays; 3D visualization; Gaze contingent displays; Foveation; Area of Interest Management; human-computer interaction; area of; interest management

Lead
Lay summary
This project aims at developing novel concepts, models and methods inspired by human visual system (HVS) and at implementing these in visualization of geographic data. When HVS's limitations are taken into consideration, it becomes evident that a great amount of detail is not perceived by humans for a variety of psycho-physiological reasons. Current computational geovisualization processes rely on well-established algorithms to remove redundant or irrelevant information to overcome hardware and/or bandwidth limitations, and much research has been done in this area. However, the biological limitations of HVS are not exploited or integrated into geovisualization displays. To understand the potential of HVS-inspired approaches, we start by looking into biological and optical foundations of fovea and the lens system, as well as stereoscopic vision. In doing this we start our work with a concept called foveation. Foveation is a computer vision method that produces a non-uniform display, typically by degrading the resolution on peripheral regions where human visual acuity is compromised by nature. These displays typically depend on having a modality (e.g., an eye tracking device, or a user-input location) to determine viewers' gaze point (point of interest (POI)) in the scene. They are often termed as gaze contingent displays (GCDs). Within the initial stages of the project a concept called geofoveation is being developed, exploring the potential of foveation in geovisualization domain (GeoF). Within GeoF, a generic model of foveation is explored for all kinds of geographic data displays. We further expand our approach by combining foveation with depth of field (DoF) simulation in a three-dimensional (3D) space for stereoscopic 3D geo-visualizations, because in this particular area data can be especially large. Within this project, we develop an extended and improved mathematical model of foveation for stereoscopic geo-GCDs based on previous research. Following this stage, a test bed will be implemented which takes both raster and vector geographic data formats as input, processes them to create gaze-dependent foveated visualizations on mono- and stereoscopic displays (GeoF II). To validate our approach, testing and comparison of the new algorithms and existing approaches will follow the implementation of the test bed. In summary, within (GeoF and) GeoF II, we focus on finding ways to better manage large geographic information visualizations, especially for low-bandwidth scenarios.

Direct link to Lay Summary

Last update: 21.02.2013

Self-organised

Communication	Title	Media	Place	Year

Title	Year

Number	Title	Start	Funding scheme

With this proposal we seek a two-year extension of a previously SNF-funded two-year project titled “Development and implementation of Geofoveation (GeoF)” (please note that the original title was “Geofoveation and Viewer Response to Foveated Displays” in the first application). The first portion of this project has an initial duration of two years (01.01.2009 - 01.01.2011). Geographic data is essentially large and complex, and on a geovisualization display often only selected features are visualized. We propose to investigate biological limits of human vision as a means to reduce the displayed information in geovisualization displays and develop methods to compress large geographic datasets by selectively removing or simplifying what is essentially not perceived by the human visual system (HVS). While the conceptual development is generic to all types of geovisualization displays and belongs to first the two years of this project (GeoF), the focus of this project will be on three-dimensional (3D) stereoscopic gaze contingent displays (GCD) and will be implemented under the framework of GeoF-II. In the proposed research we intend to continue the research that was started with GeoF, develop the concept further and implement, test and validate the proposed concepts. We conceptualize viewers’ gaze as an input modality via an eye tracking device, which provides the location of a continuously changing area of interest (AOI) over a discrete space (a geovisualization display). As the dynamic AOI moves, the display is rendered in a non-uniform manner where the center is rendered in high resolution, and periphery in gradually lower resolutions in 3D space. This rendering matches with the HVS’s natural boundaries, creating foveated GCDs. We hypothesize that efficient data management and visualization for complex geographic datasets do not only depend on cartographic, cognitive and contextual principles, but also have to consider perceptual limits of the HVS based on its biological configuration. To assess this hypothesis, our research program is based on a set of interdependent research questions, involving the development of a conceptual framework for geofoveation and supported by design, analysis and performance evaluation of implemented GCDs via a test bed (software).This project will be carried out over two years. Modest resources are required to continue employing a PhD student, to enable one collaborative visit by the student to one of the partner institutions, to further test and validate the results with an autostereoscopic display, and to support conference attendances to disseminate research findings.The results of the proposed work is firstly aimed at providing a better understanding of limitations of human vision for modeling geovisualization-specific GCDs, and secondly designing and implementing a modular software that contain novel algorithms for comparison and validation of HVS-based geographic data visualization approaches. Resulting concepts, models, methods and tools should be generic enough to be used for all types of geovisualization displays (e.g., cartographic products, 3D city models, 3D terrain models, scenario-based simulations). We expect that stereoscopic GCDs will attract attention also from other domains where stereoscopic displays and applications are relevant (e.g., surgery, films and animations, gaming).In summary, this interdisciplinary research project will have two key outcomes:•To identify the features and limitations of human visual system that can help with more efficient management and representation of geovisualizations (GeoF)•To design, develop and implement a test bed integrating multiple methods for comparative testing of biologically inspired mono- and stereoscopic visualizations for geographic data (GeoF II)