Computer Vision
Computer vision derives complex information from two-or three-dimensional images or objects through computer analysis. The technology, also called machine vision, grew from specialized research into pattern recognition. Enabling a computer to "see" is a complex process, involving many factors. In general, images to be analyzed must have any special patterns identified, then precise measurements of the characteristics of the patterns are taken and finally, a comparison of each part or pattern takes place. Jerome Lemelson was a pioneer in this field. During the 1950s, he experimented with programmed industrial robots. By equipping them with special sensors, they were able to see and could thus inspect products being passed before them. Lemelson's prototypes used light beams and photoelectric cells to scan objects, but with rapid advances in hardware and computer technology, his designs were soon improved. By 1960, he used television cameras linked to digitizer-equipped computers to measure object dimensions, detect surface imperfections, and compare colors and tolerances. While most of early vision systems dealt with two-dimensional images such as documents, Larry Roberts, a MIT graduate student, began work in the early 1960s on what was called visual scene analysis. This led to important breakthroughs in computer vision, particularly in recognizing three-dimensional scenes, a type of technology required in robotics vision. By the late 1960s, MIT, Stanford, and other universities began experimenting with robots capable of visual perception. In 1969, the Stanford Institute developed a mobile robot, which they affectionately named Shakey for its haphazard way of moving. Outfitted with TV cameras coupled to sensory feedback systems, Shakey was an excellent example of early artificial intelligence and specialized pattern-recognition technology. Another development in machine vision was the testudo (turtle in Latin), by William G.
Walter. This device was a small turtle-like object equipped with a photoelectric cell for an eye, a sensing mechanism to detect contact and motors that enabled it to turn or move forward and backwards. It could detect light sources and automatically moved toward illumination when it required recharging. One interesting tangent that arose from computer vision is computer art. An early example was produced in 1966 by Bell Lab engineers Leon D. Harmon (b.1922) and Kenneth C. Knowlton. Using a computer linked to a TV camera to scan pictures or paintings, each row in the picture is separated into points, and assigned a number representing brightness. By specifying an appropriate symbol for each number, a fairly accurate rendering of the painting is seen by the computer. The ability of electronic circuits to "see" an image has also been put to use in Israeli police work. A program called PATREC (short for pattern recognition) has been used to analyze a police artist's sketch of a criminal suspect. Running it through the main police file, similar photographs are matched and selected for further scrutiny. Computer vision has also been used in the military to provide long-distance eyes on the battlefield for military commanders and has proved valuable in analyzing satellite pictures. But perhaps the largest application of machine vision is in the automobile manufacturing field. Companies such as General Motors employ thousands of machine vision systems. As microprocessors continue to gain in performance, speed, and capacity, many innovative uses for computer vision are sure to emerge.
This is the complete article, containing 538 words
(approx. 2 pages at 300 words per page).
