Computer Pattern and Character Recognition | Research & Encyclopedia Articles

Computer Pattern and Character Recognition

Character recognition is the technology of using machines to identify symbols (usually alphanumeric characters) in order to express them as machine readable codes. The first commercial attempts at character recognition appeared in the 1950s when magnetic ink character recognition (MICR) was introduced into the banking industry. With the MICR system, characters and symbols are printed on checks or deposit and withdrawal slips using magnetic ink and are then put through a magnetizing process. Although MICR characters are readable by humans, often special MICR reader/sorters are used to interpret and sort the checks more rapidly. In other areas of industry during the 1950s, the keypunch, keytape and key-to-disk systems were being used for most data input. Although popular at that time, these devices were expensive, prone to errors, and not well suited for large quantities of data entry. The desire to improve data entry methods paved the way for the development of character recognition and optical recognition machines. Character recognition systems read numbers, letters, special characters and marks. An electronic scanning device converts the data into electrical signals and sends the signals to the computer for processing. Data may be represented in a variety of ways, including optical marks, optical characters, handwritten characters or bar codes. Optical mark recognition (OMR), which was introduced in the early 1960s, is the simplest form of character recognition. OMR consists of a series of marks, lines or bar-codes, which are detected and decoded by fixed optical scanners or hand-held wand readers. College entrance exams are commonly scored using OMR technology. Optical character recognition (OCR), a more advanced concept, is the process of optically reading typed, printed, or handwritten characters.

OCR has been hailed as the solution to computer input bottlenecks because it may help to reduce or eliminate manual keystroke operations. Early OCR machines were restricted to recognizing just one or two standardized fonts, a long-standing problem which American engineer Raymond Kurzweil (1948-) resolved to tackle. In 1974 he began developing a machine that would read to blind people. Two years later the Kurzweil Reading Machine, which could scan any type of printed page and read aloud everything from utility bills to full length novels, was enthusiastically received upon its release. Today modern OCR systems can be "trained" to recognize almost any font, as well as handwritten materials. OCR is utilized by the U.S. Post Office. OCR readers/sorters and bar-code readers began appearing in post offices in the 1960s. Using OCR technology to read and sort zip codes saves much time and effort. Computer pattern recognition developed as a specialized form of character recognition and is used by manufacturers who utilize industrial robots. When a robot picks up a part such as a screw, nut or other component, it must be able to sense the precise shape and alignment of the part in order to grasp it properly. Optical sensing gives the robot this pattern recognition ability. Although a robot can be taught to sense a physical part or object, when symbols or handwriting are the materials, the problem of pattern recognition becomes much more complex. Since the 1970s researchers have been attempting to merge pattern recognition with artificial intelligence systems. Artificial intelligence is the ability to solve thinking problems; problems which require judgment, reasoning, imagination, intuition and other human characteristics, such as emotion and loyalty. To install this kind of human intelligence in computers is the ultimate goal of pattern recognition and artificial intelligence research.