An experimental machine has been designed and constructed to be used for optical character recognition. As an integral part of the recognition system, an optical character scanner has been built which optically scans a 35mm microfilm transparency and provides binary information for the rest of the character recognition system. An IBM 1401 computer system delivers gross commands to the character recognition machine, allowing the scanner and other parts of the system to be partially autonomous. The experimental system has been used to study the problems of character recognition. It offers opportunity for considerable flexibility to do research experiments, and it is not intended to be a prototype of a commercial recognition machine.
A high-resolution, high-precision mechanical flying spot digitizer (FSD) operating on-line to an IBM 7094 computer makes available to the computer some 20,000-50,000 coordinate pairs describing a bubble chamber picture in a time of a few seconds. The computer is guided by humans in its selection of relevant data -- normally by a few rough coordinates recorded in punched cards during a pre-scan, but also by a manual over-ride at the 7094 console in the event of failure in recognition at the time of precision measurement. The system is capable of transition to purely machine recognition as computer programs acquire sufficient subtlety and reliability. The Mark II FSD at Brook-haven is capable of 10 micron resolution and 1-2 micron accuracy over a 2" x 6" area of the photographic negative. Completed systems at Brookhaven and at the University of California in Berkeley are each processing strange-particle events at about the rate of 10 conventional measuring machines, with a further expansion in rate by about an order of magnitude envisioned. Similar systems are under construction at eight other laboratories in the United States and Eurone.
A great part of research and experimental effort in the biological sciences, and indeed of most of the life sciences, is concerned with obtaining and evaluating large quantities of statistical data. Frequently the required data is obtained by visual observation and measurement through a microscope, or by study of a photographic record such as an X-ray film, a photomicrograph or an electron micrograph.
Automatic Target Recognition from aerial photographs is formulated in decision theoretic terms. Methods of representing images on photographs as multi-dimensional vectors are discussed, and the method of processing the resulting parametric representation of images by computer-aided means is presented. The electro-optical instrumentation aspects of obtaining the vector representation of photographic images is illustrated with examples.
The U. S. Naval Observatory is now in the process of constructing a 60" reflecting telescope at their site near Flagstaff, Arizona. The name of this instrument is the 60" Astrometric Reflector, and all aspects of its design and construction are directed toward very high accuracy and long-term stability. The special problem for which the Naval Observatory will use this telescope is the study of stars called Red Dwarfs, White Dwarfs, and Sub-Dwarfs. The physical characteristics of these stars differ from stars of the solar type.