Bela Julesz

Bela Julesz (19 February 1928 - 31 December 2003), also written Béla Julesz, was a Hungarian-born American electrical engineer and vision scientist who invented the random-dot stereogram. He worked at Bell Telephone Laboratories in Murray Hill, New Jersey, from 1956 to 1989, and then became director of the newly established Laboratory of Vision Research at Rutgers University. His 1960 paper "Binocular Depth Perception of Computer-Generated Patterns" used computer-generated pairs of random-dot images that look like noise to either eye alone but show a shape in depth when the two are fused. The result showed that stereopsis can work from binocular disparity alone, without any shape that either eye can recognize on its own.

Random-dot stereograms became a standard tool in vision research and in tests of stereo vision. They also fed into computational work on stereo matching and into the single-image autostereograms. In his National Academy of Sciences memoir, the vision scientist Christopher Tyler called Julesz "a founder of the field of computer vision" and "the inventor of the random-dot stereogram and its brethren". Julesz set out this work in the 1971 book Foundations of Cyclopean Perception.

Reviewed 15 October 2026. Checked birth/death dates, education, Bell Labs and Rutgers career, honors, the 1960 BSTJ paper details, the 1965 Howard Wise show, legacy claims and all journal citations against the Spatial Vision, APS, NAS and PLoS Biology obituaries, the original papers, Crossref and the 2023 TVST abstract. About review dates.

Early life and education

Julesz was born in Budapest on 19 February 1928. According to the memoirs by Thomas Papathomas and Christopher Tyler, he took private mathematics lessons from Rózsa Péter, the founder of recursive function theory. He received a diploma in electrical engineering from the Technical University of Budapest in 1950 and was an assistant professor in the Department of Telecommunication in 1950-51. From 1951 to 1956 he worked as a research engineer in microwave systems at the Telecommunications Research Institute in Budapest. In 1956 the Hungarian Academy of Sciences awarded him a Ph.D. for the thesis "Study of TV Signals with Correlation Methods". Tyler's memoir adds that an earlier dissertation on radar methods was classified by the military as soon as it was accepted, so Julesz had to write the second one on television bandwidth compression.

He married Margit Fasy on 7 August 1953. After the defeat of the 1956 Hungarian Revolution the couple left Hungary by crossing a river at the border and settled in the United States, where Julesz was naturalized in 1963.

Career

Bell Laboratories

Julesz joined Bell Telephone Laboratories in Murray Hill, New Jersey, as a member of the technical staff in 1956. Papathomas writes that he devoted his full time to visual research after 1959, especially depth perception and pattern recognition. Ralph Siegel's obituary in PLoS Biology relates Julesz's own account of how this began: he was assigned to test random binary sequences at Bell Labs, and he plotted them as rows of an image so that the human visual system could spot any repeats or spatial correlations. He headed the Sensory and Perceptual Processes Department from 1964 to 1982 and the Visual Perception Research Department from 1983 to 1989. He retired from Bell Labs in January 1989 after 32 years.

During his Bell Labs years he took frequent sabbatical leaves at universities. He was visiting professor of experimental psychology at the Massachusetts Institute of Technology in 1969, where his lectures became the basis of his first book, and he visited the University of Western Australia in 1973. In 1975-76 he taught at ETH Zurich and the neurology department of the University of Zurich, where he developed a method with D. Lehmann for testing stereopsis in human infants. He was a Fairchild Distinguished Scholar at the California Institute of Technology for the winters of 1977 to 1979 and again in 1987, working in Derek Fender's laboratory on binocular retinal stabilization and with John Allman on visually evoked potentials in monkeys. He was Regents Lecturer at the University of California, Berkeley, in 1983.

Random-dot stereograms

Julesz's 1960 paper in the Bell System Technical Journal (manuscript received 31 March 1960, published in the September 1960 issue) asked whether the brain matches the two eyes' images by first recognizing patterns in each eye's view, or by first combining the two views and then searching the fused result. To separate binocular disparity from every other depth cue, he had an IBM 704 computer generate image pairs of about 10,000 picture elements, each assigned a random brightness level. The surround was identical in the two images, while a square region in the center of one image was shifted horizontally. Viewed with one eye, each image looked completely random; viewed stereoscopically, the square appeared clearly in front of or behind the surround. The pictures were produced through a television scanner developed at Bell Labs, and the issue included a pair of Fresnel lenses so readers could view the stereo illustrations.

The paper concluded that depth could be seen with no monocular cues or Gestalt, and suggested that the properties needed might be simple enough to simulate with computers, so that the findings "might give a new impetus to the development of devices that will determine depth automatically". Siegel notes that the Journal of the Optical Society of America rejected Julesz's first report on random-dot stereograms and later published his second paper. His paper "Stereopsis and Binocular Rivalry of Contours" appeared in that journal in 1963. Siegel places the work in the history of the stereoscope, which Charles Wheatstone demonstrated in 1838, and of Wheatstone's debate with David Brewster over how depth is perceived; he writes that the 1960 experiment showed that "stereoscopic depth could be computed in the absence of any identifiable objects". Julesz followed it with "Binocular Depth Perception without Familiarity Cues" in Science in 1964.

He later extended the method to dynamic random-dot stereograms and random-dot cinematograms (moving sequences), which Papathomas writes "revolutionized research in stereoscopy and motion perception". The Spatial Vision obituary credits Julesz with finding hysteresis in binocular depth and monocular movement perception, which led to cooperative models of depth and motion perception.

Foundations of Cyclopean Perception

Julesz wrote Foundations of Cyclopean Perception (University of Chicago Press, 1971) from his 1969 MIT lectures. Tyler writes that Julesz called the ability to see form only through binocular depth "cyclopean vision", taking the term from Ewald Hering's 1868 idea of a single metaphorical eye that combines the information from both eyes. In Siegel's summary, the book argued that the images from the two eyes are combined early in visual processing into a single view that carries depth information, before motion, color and contrast are analyzed. Papathomas records that the book was included among the hundred most influential works in cognitive science of the 20th century by the University of Minnesota's Millennium Project. Julesz wrote a new preface for a reprint a few weeks before his death, and MIT Press reissued the book in May 2006.

Julesz also coined the term "psychoanatomy" for tracing the flow of information through the stages of the visual system with psychophysical stimuli such as random-dot stereograms.

Texture perception

Julesz also studied how people tell visual textures apart, using computer-generated textures with controlled statistics. In 1962 he published a formal conjecture that textures differing only in their higher-order statistics (agreeing up to second-order statistics) could not be discriminated within the short time scale of pre-attentive vision. Tyler writes that the conjecture stood until 1975, when Julesz published the first textures that contradicted it, and that this led to his theory of textons, which Julesz regarded as his greatest contribution to science. He presented the theory in "Textons, the elements of texture perception, and their interactions" in Nature in 1981. With Dov Sagi he studied attention, including the 1985 paper "Fast noninertial shifts of attention".

Computer art

Julesz's article "Texture and Visual Perception" appeared in the February 1965 issue of Scientific American, with one of his color patterns on the cover. According to his Bell Labs colleague A. Michael Noll, the gallery owner Howard Wise saw that cover and invited Julesz to exhibit; Julesz asked Noll to join him. Their show "Computer-Generated Pictures" ran at the Howard Wise Gallery in New York in April 1965, with about eight two-dimensional works by Julesz, including the random patterns "Labyrinth" (1960) and "Kaleidoscope" (1962). Papathomas and Tyler describe it as the first computer art exhibition in the United States. Tyler writes that Julesz did not claim the pictures as art and said they were shown only for their scientific interest.

With Leon Harmon he made the block portrait of Abraham Lincoln used in their 1973 Science paper on masking in visual recognition. Tyler recounts that Salvador Dalí later used one of Julesz's random-dot stereograms, of a torus, in one of his works, and Papathomas writes that Dalí consulted Julesz about producing a modified version of one of his stereograms.

Rutgers University

In 1989 Julesz became a State of New Jersey Professor of Psychology and director of the newly established Laboratory of Vision Research at Rutgers University. He later retired from Rutgers as professor emeritus. His second book, Dialogues on Perception, was published by Bradford Books/MIT Press in 1995; it is written as dialogues in which Julesz questions his own scientific work. He died on 31 December 2003, 47 years to the day after he had started at Bell Laboratories. At the time he was working on a second edition of Foundations and a new book, The Enjoyment of Vision by Eye and Intellect.

Honors

Year	Honor
1980	Elected Fellow of the American Academy of Arts and Sciences
1982	Corresponding member, Göttingen Academy of Sciences; Neurosciences Associate of the Neurosciences Research Program at Rockefeller University
1983	MacArthur Fellowship (class of February 1983)
1983	Honorary member, Hungarian Academy of Sciences
1985	Dr. H. P. Heineken Prize, Royal Netherlands Academy of Arts and Sciences
1987	Elected member of the National Academy of Sciences
1989	Karl Spencer Lashley Award, American Philosophical Society; Fellow of the Society of Experimental Psychologists
1995	Elected member of the American Philosophical Society

He was also a fellow of the American Association for the Advancement of Science and of the Optical Society of America, chaired the latter's Vision Group in 1980-81, and served on the editorial boards of Perception, Biological Cybernetics and Spatial Vision. The MacArthur Foundation describes him as an experimental psychologist, educator and communications engineer who pioneered computer-generated textures with controlled statistical and geometric properties. He held four US patents, two of them on quality inspection of VLSI components.

Legacy

Random-dot stereograms contain no depth cue other than disparity, which made them a tool for isolating stereo vision. The Spatial Vision obituary describes the technique as widely used in research on depth and motion perception and as a test of stereopsis in human infants and monkeys. Julesz worked on diagnosing stereoblindness in strabismic infants by measuring visually evoked potentials with dynamic random-dot stereograms. A 2019 review in the journal Vision by Papathomas, Kazunori Morikawa and Nicholas Wade notes that his contributions to stereopsis led to new methods for studying disparity-tuned neurons in the early visual cortex.

In computer vision, the random-dot stereogram became a test case for automatic stereo matching. David Marr and Tomaso Poggio's 1976 paper "Cooperative Computation of Stereo Disparity" derived an algorithm for the correspondence problem and showed that it "successfully extracts information from random-dot stereograms". Tyler's memoir links Julesz's sabbatical at MIT with his contact with this computational vision community. Julesz himself had published "Towards the automation of binocular depth perception (AUTOMAP)" at the IFIP Congress in Munich in 1962.

Magic Eye, the maker of single-image autostereograms, states that the engineer Tom Baccei and the 3D artist Cheri Smith built on Julesz's research, and credits him with the first black-and-white random-dot stereogram pair. The company dates the single-image method to Christopher Tyler and the programmer Maureen Clarke, after Julesz's work.

Random-dot stereograms are still used to measure stereoscopic depth perception, including in studies that compare them with newer stereotests for virtual reality headsets. A 2023 study in Translational Vision Science & Technology by Sylvie Denkinger and colleagues described eRDS v6, a dynamic random-dot stereogram test, and version 2 of the Vivid Vision Stereo Test, a local or "contour" stereotest for VR headsets. In tests with 64 subjects, eRDS v6 measured small thresholds (below 10 arcsec), while the VR test did not measure stereoacuities below 15 arcsec.

Selected publications

Year	Title	Venue
1960	Binocular Depth Perception of Computer-Generated Patterns	Bell System Technical Journal, vol. 39
1963	Stereopsis and Binocular Rivalry of Contours	Journal of the Optical Society of America, vol. 53
1964	Binocular Depth Perception without Familiarity Cues	Science, vol. 145
1965	Texture and Visual Perception	Scientific American, vol. 212
1971	Foundations of Cyclopean Perception (book)	University of Chicago Press; reissued by MIT Press, 2006
1981	Textons, the elements of texture perception, and their interactions	Nature, vol. 290
1995	Dialogues on Perception (book)	Bradford Books/MIT Press

See also

• Stereopsis
• Binocular disparity
• Stereoscopy
• Binocular vision