| Every day we move through a sea of invisible information. As computation, sensing, and display become more mobile and distributed, interaction shifts from our desktop to our environment. This shift changes how we interact with our surroundings, creating the opportunity for situated visualization ---visual representation of data presented in its spatial and semantic context. This dissertation contributes novel interaction and presentation techniques for situated visualization in three research areas: mobile visualization, objects as context, and scenes as context. In support of this research, we make further contributions by developing algorithms, architectures, and working artifacts.;For mobile visualization, we present results from a field study and task analysis of botanists performing species identification in the field. We develop an iteratively designed, extensible Electronic Field Guide (EFG) system and architecture, a conceptual data model, and interface (LeafView) for mobile visualization. Field experiments show improved identification speed and interaction efficacy.;Next, we explore spatially- and semantically-driven situated visualization using objects as context in head-worn augmented reality (AR). We develop and evaluate tangible AR and head-movement controlled AR interaction techniques. In interviews following lab experiments, participants reported improved speed for inspection and comparison and a preference for tangible AR. Building on this work, we design, develop, and evaluate visualization and activation techniques for discovering and learning gestures for these user interfaces ( Visual Hints). Lab experiments show preference for visual hints that combine overlaid graphics and animation. In addition, we investigate menu techniques, activated by shaking an object, for interacting with visualizations (Shake Menus). We compare display-, object-, and world-referenced coordinate systems for presentation of menu options. Lab experiments show increased speed and accuracy when using the display-referenced coordinate system.;Finally, we present techniques for spatially-driven visualization in the context of physical scenes. Based on our field study of site visits by urban designers, a hand-held AR visualization tool (SiteLens) embodying these techniques enables interaction with invisible aspects of urban sites, such as georeferenced sensor data. Field experiments provide evidence for new insights derived from situated visualization, preference for specific representations, and improved interaction with data using a novel stabilization algorithm. |
