single-rb.php

JRM Vol.9 No.2 pp. 83-84
doi: 10.20965/jrm.1997.p0083
(1997)

Editorial:

Special Issue on New Evolution in Vision System

Masanori Idesawa

Professor, Graduate School of Information Systems, The University of Electro-Communications, 1-5-1, Chofugaoka, Chofu-shi, Tokyo, 182, Japan

Published:
April 20, 1997

A human being can carry on his activities flexibly in his three-dimensional environment by grasping and judging objects and various situations in the external world through his senses. Of these senses, the visual sense plays the most important role. And almost all the processing of visual information is carried out by the brain, just as the saying, “The eye is a branch of the brain,” goes. We see things not with the eye but rather with the brain. The mechanism of information processing inside the brain is an ultimately fine example of an information system, and in recent years, attention has increasingly been paid to the elucidation of this mechanism among information scientists as well as information processing engineers. In the past, studies on the visual function have been carried out and developed energetically by neuro-physiologists and anatomists, as well as by psychologists and psycho-physicists. In recent years, new research techniques and measurement methods have been developed and refined and, so, past knowledge has been corrected or new discoveries made; for this reason, it is now _ cnsidered necessary to carry out studies from new standpoints, including ng information science and other scientific fields. It is desired that automatic systems, such as robots and the like, be also equipped with a visual system corresponding to the human visual sense. Studies on the artificial realization of visual functions and their applications, which started roughly in the latter half of the 1960s, were concerned, at the initial stages, with the reading of characters and figures as pattern recognition and image processing. After that time, various fields of research have been formed and developed, such as image understanding as well as robot vision and computer vision as visual systems for intelligent robots. From about 1985, research called “Shape from X” based on computational vision by Marr1) has been carried out actively. Recently, however, the research in this field has collide with a mysterious wall, and has therefore been somewhat in a stagnant state. For this reason, a new breakthrough is strongly desired just about now. In order to make clear the true nature of this wall and bring about a breakthrough, this special issue concerning “new development of vision systems” has been planned with the view to constructing a new paradigm in the artificial realization of visual functions, such as computer vision, etc., and their applications. It is our wish that progress be made not only in the construction of a new paradigm for promoting themes and studies in robot vision, stereo vision, etc., including recent knowledge and simulation of the human visual system, but also in research and development work with future research advances taken into account. Now, the present author would like to reminisce about research situations in this field at about the time when he was a student. In the latter half of the 1960s, the automatic reading of marks and characters under the name of pattern recognition was being pushed forward as a national project. It was around that time that an automatic zip code reading system was created. And, attempts were started on image processing and on intelligent robots furnished with a visual capability. A little before that time, efforts had been seen to learn and apply the mechanisms of living creatures in engineering fields, such as cybernetics, bionics,etc., with the result that studies using animals have been carried out in earnest. At that time, a large number of cats, rabbits, and others were forced to cooperate (?) in a sacrificial and devoted way as “volunteers(?)” in studies by engineers. At about the time when the enthusiasm for this bionics cooled off, the present author as a graduate school student was fortunate enough to pay a visit at national research institute which was playing a leading role in research on robots and pattern recognition in Japan. An image-acquiring device based on a TV camera, which can be obtained cheaply nowadays, was priced at that time at more than several million yen in spite of its slow input speed (taking a dozen or so minutes for inputting a single image) and low performance. At that time, there were only a handful of universities and research institutes which were in possession of an image-acquiring device using a TV camera. For this reason, a stereo vision system equipped with a pair of TV cameras was considered a dream beyond a dream. The research who was giving a tour at that time said by way of an explanation, “If another TV camera image acquiring device were available, it would be possible to realize a stereo vision system capable of recognizing three-dimensional objects through binocular stereo viewing, but we have been able to purchase only a single unit due to a low budget and a high cost of the unit.” At that time, the author, who was also interested in stereo vision in connection with the development of “a system for generating solid figures from three-dimensional drawings”2), had been convinced that binocular stereo viewing was not such a simple topic. For this reason, this arrogant student was audacious enough to ask, “I think that the use of two TV cameras will make possible the input of a pair of two-dimensional pictures but will not lead to the realization of stereo vision, so do you have any new idea?” Incidentally, it is now possible to obtain a device that costs only a 200th of the price charged at that time but has a level of performance (input speed) more than 200 times as high as before. In the case of computers, too, the performance has improved dramatically, and the price has become remarkably cheaper. From the situation as existed at that time, today’s situation appears totally to be a dream, and is one in which stereo vision would be a simple matter. What is surprising, however, is that unfortunately stereo vision has not been so simple as was thought at that time and is far from being realized even now. And, as stated previously, studies in this field have recently crashed an inexplicable wall, and are somewhat in a stagnant state. A new breakthrough is desired at about this time, and it is getting increasingly more opportune to make use of results achieved in fields, such as psychology, brain science, and the like. This is by itself a very good thing, but blind adoption may be somewhat disputable. It is not the case that all the past knowledge in these fields is totally correct, and a great deal of that knowledge may have to be corrected for studies to be conducted from a new point of view. In introducing such results, it is strongly desired to review the current situation from a new standpoint; in other words, rather than repeating the age of cybernetics or bionics of that time, the present age is hoped to become an age of “neo-cybernetics” or “neo-bionics” in the true sense of the word.

Cite this article as:
Masanori Idesawa, “Special Issue on New Evolution in Vision System,” J. Robot. Mechatron., Vol.9, No.2, pp. 83-84, 1997.
Data files:

*This site is desgined based on HTML5 and CSS3 for modern browsers, e.g. Chrome, Firefox, Safari, Edge, Opera.

Last updated on Sep. 24, 2021