Showing papers on "Optical character recognition published in 1971"

Segmentation Methods for Recognition of Machine-printed Characters

Abstract: This paper reports an investigation of some methods for isolating, or segmenting, characters during the reading of machine-printed text by optical character recognition systems. Two new segmentation algorithms using feature extraction techniques are presented; both are intended for use in the recognition of machine-printed lines of 10-, 11- and 12-pitch serif-type multifont characters. One of the methods, called quasi-topological segmentation, bases the decision to “section” a character on a combination of feature-extraction and character-width measurements. The other method, topological segmentation, involves feature extraction alone. The algorithms have been tested with an evaluation method that is independent of any particular recognition system. Test results are based on application of the algorithm to upper-case alphanumeric characters gathered from print sources that represent the existing world of machine printing. The topological approach demonstrated better performance on the test data than did the quasi-topological approach.

An Interactive System for Reading Unformatted Printed Text

Abstract: A system intended to provide input of printed text to computers is applied to published patents, annotated law reports, and technical journals.

Some Studies in the Interactive Design of Character Recognition Systems

Abstract: A project to develop an effective tool for the interactive design of optical character recognition systems is described The initial design system and several techniques and experiments to improve the system are also reported Results of an experiment to design a recognition system for typewritten characters are included The recognition logic required 113 man-hours to design and recognized over 999 percent of its test characters It is also small and can run at high speeds on standard digital computer systems The results of several experiments and recommendations for future experiments to extend these results to hand-printed characters are also included The development of more powerful multilevel logic forms and suggestions for making these forms easy to use in an interactive design environment are particularly important results of these studies

Feature Extraction in an Optical Character Recognition Machine

Abstract: A feature extraction technique for the next generation of optical character readers must be able to rapidly generate a quantity of discriminant functions using a minimum of equipment. The procedure described in this note generates a series of reference patterns for two-class discrimination and a feature ordering tree which may also be used to decode the sample class. Special hardware is used to provide practical learning times for handprinted characters.

Pattern recognition technique

Abstract: Technique operates regardless of pattern rotation, translation or magnification and successfully detects out-of-register patterns. It improves accuracy and reduces cost of various optical character recognition devices and page readers and provides data input to computer.

Line finding system and method for character recognition

Abstract: In an optical character recognition system the position of a line of characters is detected during the period when the recognition system is not reading the characters but is executing an aging scan which is normally used only for moving the recognition beam around to prevent burns on the face of the recognition system video scanner. The line positions are stored for use during the read mode of the recognition system. The positions of lines actually read during the read mode are compared with these stored positions to determine that all lines in stored positions were actually read. If not read, the read mode is re-executed to read these lines.

The IBM 1275 Recognition System and Its Development

Abstract: Magnetic Ink Character Recognition (MICR) was developed about 10 years ago for the automation of cheque and transfer processing in banks and Postgiro’s. In the U.S.A. the E-13B type fount was developed and in Europe CMC-7; their general characteristic is eight characters per inch printing pitch, special printing equipment and a fairly high degree of print quality control. In continental Europe, MICR has not been applied extensively and there is a strong trend noticeable towards Optical Character Recognition (OCR). The primary reason for this is a ten character per inch printing pitch which allows for the use of regular printers, e.g. high-speed printers. It is felt that OCR can provide a better balance between document preparation and reading costs. The subject recognition system was developed specifically with this reasoning in mind. Details of this development are given below.

Design of an Experimental Laboratory for Pattern Recognition and Signal Processing

Abstract: An interactive computer-controlled scanning and display system has been in operation at the IBM T J Watson Research Center for two years The system includes two flying-spot scanners, specially interfaced to a process control digital computer, dot-mode and vector displays, analog input and output facilities, and a variety of other experimental equipment The system design and programming support are described and typical applications in scanner control, optical character recognition and image processing are presented

Character recognition and visual processing

Abstract: An optical character recognition system has been developed which simulates some of the functions of visual physiology and psychology. Neurophysiology has recently demonstrated that lateral inhibition occurs throughout the visual system and produces operators for feature extraction. Psychological evidence suggests that the visual features are used in a hypothesis generation and test procedure. The model for these processes is presented within a linguistic frame-work. A learning procedure is also described which constructs structured descriptions of distinctive features for co-classed characters.

Automatic typeset input techniques evaluation

Abstract: : The need to design multi-font print readers is becoming critical for the input conversion activities of the Air Force Foreign Technology Division Machine Translation facilities. This effort was to evaluate existing optical character recognition capabilities toward the total requirement of a Russian typeset print reader. In this research four pages of original scientific Russian text were used as the data base. The contractor demonstrated that the scanning and conversion of Russian text by OCR is feasible and potentially economical. The second objective was to identify the problem areas and compile solutions to them. The results of the analysis of scanning the four samples indicate that secondary recognition and increased digital resolution will be most effective in reducing the total error rate. An ultimate total error rate of less than .5% appears achievable. (Author)