GB1129572A - Improvements in or relating to character recognition systems or apparatus - Google Patents

Abstract

1,129,572. Character recognition. SCANDATA CORP. 11 April, 1967 [21 April, 1966], No. 16480/67. Heading G4R. In a character recognition system, a computer edits text machine-read from a document on which editing characters are present in machinereadable form, in response to indication of character identity and location. Editing corrections to printed text on a document are made by writing the appropriate editing characters under the locations in the lines of print where correction is required, the possible editing characters having a vertical bar and a selection from six possible horizontal bars, the latter projecting from the top, middle and bottom of the vertical bar to the left and right respectively. The flying-spot scanner of a character recognition system performs a raster scan of each text character in a given line, then of any editing characters below the line, then of each text character in the next line &c., the resulting video signals being digitized and stored in a shift register capable of holding all the signals from one character. Threshold logic fed from the register recognizes character features, and patches of background to check registration, producing signals which feed AND and OR logic to recognize the character. The character identity is coded and sent to a computer together with the location of the character. By comparing the locations of text characters and editing characters the computer discovers where editing is required and performs it in accordance with the editing character identies. The meaning of a given editing character may depend on its location or the adjacent editing characters, as well as on its own shape (identity).