Head Tracked Auto-Stereoscopic Displays

Head Tracked Auto-Stereoscopic Displays

Philip Surman (De Montfort University, UK)
DOI: 10.4018/978-1-4666-4932-3.ch005


This chapter covers the work carried out on head tracked 3-D displays in the past ten years that has been funded by the European Union. These displays are glasses-free (auto-stereoscopic) and serve several viewers who are able to move freely over a large viewing region. The amount of information that is displayed is kept to a minimum with the use of head position tracking, which allows images to be placed in the viewing field only where the viewers are situated so that redundant information is not directed to unused viewing regions. In order to put the work into perspective, a historical background and a brief description of other display types are given first.
Chapter Preview

Historical Background

Historically, 3D display systems were often considered an obscure research interest whose fruits were of dubious practical or commercial value. Recent technological advances, coupled with the demands of sophisticated methods of interaction (for example virtual reality) have reawakened interest in the techniques and applications of 3D imagery. So we have to consider the terminology, as in most discussions of 3D displays. Historically, authors have often used many definitions and descriptions ambiguously, often serving only to confuse the reader. For the purpose of this discussion a 3D display system is defined as one that provides additional information to the observer as a consequence of the observer’s possessing stereoscopic vision. In order for any technological advancement to gain widespread acceptance it should be capable of something new and previously unachievable, or perform its task so much better than the existing alternative that it naturally supersedes it.

Our primary interest in stereoscopy involves its application to eliciting depth and detail in an otherwise flat image. However, there are many other applications, including the detection of counterfeit currency and documents, and the investigation of convergence disorders in ophthalmology, to name two important examples. It seems natural to utilise stereoscopy in the display of electronic imagery. We see the world stereoscopically, so presumably we should endeavour to recreate the visual experiences to which we are accustomed in other images. This argument was no doubt applied when colour was introduced into electronic imagery, although the universal acceptance of colour was achieved with some opposition. Indeed, at its inception colour TV was considered by many as an extravagance, receivers costing as much as a small family car. Nowadays most households own more than one colour TV receiver.

The visual perception of depth depends upon image disparity. Our eyes are horizontally disposed and receive disparate images. A synthetic image must include that disparity. This is generated in computer graphics by geometric transformations. However, when viewing the world we are also receptive to other kinds of disparity. The surface reflectance of some materials can result in the images formed on each of the retinas differing in colour and brightness. Binocular mixture of the hues occurs, and a physiological phenomenon known as retinal rivalry accepts and rejects the binocular mixture in rapid alternation, and as a result imparts a lustre-like characteristic to the perceived image of, for example, metals and crystals. Although it may be useful if a 3D display system can portray a likeness of such materials more realistically, the effect may appear as a defect. Points of correspondence in computer-generated images may appear to be differently coloured, and this can give rise to a distracting 'twinkling' at these points.

Binocular vision is only one member of a set of properties commonly referred to as ‘visual depth cues’. These properties may be psychological or physiological or both (Graham, 1965). These cues include: static and dynamic parallax, linear and aerial perspective, size constancy, accommodation, and, of course, stereopsis. An examination of these terms is worthwhile, particularly for those involved in computer graphics, where the objectives may be clear but expense dictates a compromise. For example, given the choice between stereoscopic wire-frame images and monoscopic shaded images, which should we use? There is no clear answer, but an understanding of the factors involved may be useful.

Imparting the illusion of depth to a two-dimensional image can hardly be claimed to be a new idea. For hundreds of years artists have been familiar with the techniques. However, there are two aids to depth perception that have not been fully exploited, particularly in the field of computer graphics, namely motion effects (dynamic parallax) and stereopsis.

Although not possible in a still picture, dynamic parallax can be readily represented in modem animated art. For example, any depth ambiguities between two objects are resolved if one object moves in front of another and thus obscures the view of the latter. This effect is used in computer-generated images but is usually limited. In a real-life situation the effect is present at all times, from the relative motion of both the viewer and the object viewed. Stereopsis has been employed with varying success in computer graphics, the graphic arts, and in television systems, but while the effect is well understood it can be fairly difficult to implement. The exploitation of binocular and motion depth cues is the goal of many 3D displays, and there have been a number of approaches.

Complete Chapter List

Search this Book: