More Augmented Reality X-Ray Systems

Augmented reality will make superheros of us all.  A preview of what you can expect for ISMAR2010 from Magic Vision Lab.

Sandor, C., Cunningham, A., Dey, A., and Mattila, V.
An Augmented Reality X-Ray System based on Visual Saliency
To appear in: Proceedings of IEEE International Symposium on Mixed and Augmented Reality, Seoul, Korea, October, 2010.

In the past, several systems have been presented that enable users to view occluded points of interest using Augmented Reality X-ray visualizations. It is challenging to design a visualization that provides correct occlusions between occluder and occluded objects while maximizing legibility. We have previously published an Augmented Reality X-ray visualization that renders edges of the occluder region over the occluded region to facilitate correct occlusions while providing foreground context. While this approach is simple and works in a wide range of situations, it provides only minimal context of the occluder object.

In this paper, we present the background, design, and implementation of our novel visualization technique that aims at providing users with richer context of the occluder object. While our previous visualization only employed one salient feature (edges) to determine which parts of the occluder to display, our novel visualization technique is an initial attempt to explore the design space of employing multiple salient features for this task. The prototype presented in this paper employs three additional salient features: hue, luminosity, and motion.

We have conducted two evaluations with human participants to investigate the benefits and limitations of our prototype compared to our previous system. The first evaluation showed that although our novel visualization provides a richer context of the occluder object, it does not impede users to select objects in the occluded area; but, it also indicated problems in our prototype. In the second evaluation, we have investigated these problems through an online survey with systematically varied occluder and occluded scenes, focussing on the qualitative aspects of our visualizations. The results were encouraging, but pointed out that our novel visualization needs a higher level of adaptiveness.