Saturday, 2 November 2013

A history of research on 3-D vision

Department of Experimental Psychology, University of Oxford, UK
Ideas about our perception of the 3-D world can be traced back to the writings of the ancient Greeks but a more useful starting-point for considering recent advances in the field might be Berkeley's (1709) claim that “distance, of itself and immediately, cannot be seen”. This led Helmholtz to propose the idea of “depth cues” and to make the distinction between two sources of information about depth and distance: (i) information that leads to an actual perception of depth (including accommodation, vergence, motion parallax and binocular disparities) and (ii) information that “merely enables us to form some idea as to distance”. These were Helmholtz's primary and secondary cues. Moreover, he regarded the use of secondary cues, such as perspective and shading, as crucially dependent on learning and experience. The 19th century invention of the stereoscope and the mid-20th century development of random dot techniques for studying both binocular stereopsis and motion parallax have provided us with unequivocal evidence that human observers are able to use these two sources of information in isolation. Moreover, research has shown that both disparities and motion parallax allow us to discriminate between targets that are separated in depth by less than 5–10 arc sec. Over the last twenty years, the combination of mathematical analysis and experimental research has shown that human observers are also sensitive to the small differences in the vertical positions of images in the two eyes that provide information about the absolute distance to objects in the world. However, research has also shown that the situation is not quite as straightforward as Helmholtz suggested. Not only do the primary cues of accommodation and vergence play a rather limited role in 3-D perception but it has also been shown that the perspective information that Helmholtz believed merely enabled us to form an “idea” of distance can completely overrule binocular disparities when the two cues are put into conflict. In this lecture, I would like to describe some of the important perceptual, physiological and computational discoveries in the field of 3-D vision over the past decades and to show how the use of converging evidence has advanced the field. In addition, I would also like to discuss some of the important conceptual issues – the idea of depth cues, the usefulness of looking as depth cues in isolation and the merits of using naturalistic rather than artificial stimulus situations – as pointers to the directions research might take in the future.
Grant support: University of Oxford

0 comments:

Post a Comment