Abstract
This study evaluates the technique that can present virtual objects at freely-selected distances of tens to hundreds of meters in optical see-through mixed reality. Presenting virtual objects at freely-selected such far distances is crucial to outdoor applications of optical see-through mixed reality, e.g., outdoor navigation and outdoor annotation. However, it is difficult for conventional techniques because they control virtual objects’ depth perception with binocular disparity that does not work at the great distances. In the proposed technique, depth cues separated into three hierarchies are given virtual objects for each hierarchy. In the first hierarchy, physiological cues, e.g., vergence and binocular disparity, are given to indicate virtual objects are not at near depths. In the second hierarchy, pictorial cues, e.g., familiar sizes and luminance contrast, are given to indicate rough depths of virtual objects. In the third hierarchy, cues based on scene understanding are given to indicate accurate depths of virtual objects. In the evaluation, the effect of placements, which are one of cues based on scene understanding, are examined with virtual objects superimposed into stereo photographs. Their perceived depths are measured, and the results indicate that accurate depths of virtual objects can be indicated with placements of virtual objects. Such findings support the proposed technique, and we demonstrate its feasibility.
