Random dot
stereograms

Random dot stereograms are composed of two arrays (or frames) of randomly scattered dots. The arrays are identical except for a certain area(s) in one of the arrays which has been horizontally shifted to introduce disparity between the two arrays. The two arrays (images) are usually displayed side by side to allow the observer to compare them. When viewed monocularly the viewer is confronted with a mass of dots and fails to make any correspondences across the images. However when the images are stereoscopically fused (by crossing the eyes then either converging or diverging them - try to gaze beyond the images) correspondences across dots are made and the shifted area(s) in the stereogram appears at a separate depth level from the unshifted area. Figure 4, below, shows a stereogram example which when fused should reveal a square area in the middle of the image at a different depth level.

Figure 4
Figure 4: A random dot stereogram


Figure 5: The shear surface of the stereogram

Stereograms are best computer generated. Not only are larger examples easier to produce, but monitoring dot densities and creating image disparities proves much more accurate and straightforward. For successful testing, statistical information must be easily obtained and accurate, and with the stereogram this is possible. The main point about the stereogram is that it doesn't provide the viewer/computer with any high level cues (information) as to what is being viewed. This means that it is the viewer's ability to stereoscopically fuse two images that is solely under test. The monocular and binocular stages of the visual process appear distinct which allows the researcher to study carefully the effects of both stages. Finally stereograms are very efficient test tools. Every dot has the potential to be matched, so large amounts of disparity information are made available.


[ The stereo correspondence problem | Panum's fusional area ]

Comments to: Sarah Price at ICBL.