JPS6364314B2 - - Google Patents

Description

[Detailed Description of the Invention] The present invention reads characters from a manuscript using a kanji OCR (optical character reading) device, recognizes and stores the characters, provides the data to a computer typesetting machine, and then creates a printing block. Regarding the printing kanji OCR device.

Conventionally, when performing computerized typesetting on this type of device, the input of kanji data has relied on manual operations, but this method requires a great deal of effort and skill. There were limits to the improvement in efficiency through learning.

The present invention has been made in order to solve the bottlenecks of these artificial input means, and its purpose is to provide a device that automatically performs everything from the input of reading a manuscript to the output of a printing plate. purpose.

FIG. 1 is a block diagram of one embodiment of the present invention.

9 represents the printed kanji OCR device, 1 is the original before being read, 2 is the original after being read, 3 is the original
A CRT display and a keyboard, 4 a tablet cursor, 5 a mechanical scanning section, 6 a preprocessing/recognition section, 7 a core memory, and 8 a control device.

10 is the stored character code, 11 is the stored character size and position data, 12 is the editing machine, 13 is the magnetic tape, 14 is the font memory, 15 is the arithmetic unit (CPU), and 16 is the page memory. , 100 is a laser beam printer, 17 is a galley, 18 is a printing plate, and 19 is a layout document supply system.

The printed kanji OCR device 9 reads any part of the document/original 1 supplied by the operator and outputs the result via a storage medium such as a magnetic tape.

It is composed of a mechanism scanning section 5, a preprocessing recognition section 6, a dictionary pattern memory 7, and a control device 8 as main members.

The mechanical scanning unit 5 includes a CRT display and a keyboard 3 for specifying an arbitrary part of a document as a reading area through dialogue (format specification);
A tablet and cursor 4 for inputting coordinates are also attached.

The reading paper/original 1 transport section transports the original 1 with a thickness of 40 kg or more under the reading head while vacuum suctioning it to the reading table. The paper feed pitch can be freely changed in units of 50 μm according to instructions from the control device 8.

The scanning/photoelectric conversion section is a CCD (charge coupled
(charge-coupled device) A reading head incorporating a self-scanning element together with a lamp and an optical system is driven at a constant speed of 30 cm/sec by a servo motor to perform scanning and photoelectric conversion in the row direction. This part has a resolution of 60 μm and can sample 9-point type at 45 points horizontally and vertically. The inclination of character lines is allowed up to ±3°, and the scanning range is 20 cm/line.

The first preprocessing part (detection cutting, sampling) of the preprocessing/recognition part 6 uses a detection first appearance method to predict the character position in one line, detect and correct irregularities in character arrangement, After cutting out the characters in one line, a sampling operation is performed. The second candidate character selection parameter extraction section calculates a complexity index from the detected and extracted character pattern, and extracts a four-sided code from the sampled pattern. The third candidate character list creation section uses these two parameters as keys and uses dedicated hardware to quickly select candidate characters for the input pattern from among all characters to be recognized.
The third similarity calculating section calculates the composite similarity or mixed similarity for the characters given in the candidate character list from among the 2000 character dictionary patterns. Note that the fourth part is a candidate character list creation section. The core memory 7 of the dictionary pattern memory is 1 megabyte of core memory. Dictionary patterns can be freely selected and rearranged by the configuration program stored in this part, and a plurality of fonts can be mixed.

The control device 8 operates each part of the hardware in a pipeline, determines the reading result from the similarity value, and outputs it to the magnetic tape, and the character recognition speed is 100 characters/second. For characters that cannot be read, the first to fifth characters are selected in order of similarity to the scanning pattern.
Display on CRT display. Through dialogue, you can receive phonetic input from the operator and modify the answer letters.

In this way, in the control device 8, the character code punched on the tape 10 is determined by the size, position, It is input to the editing machine 12 along with various data related to the layout, such as top, bottom, left and right margins, horizontal and vertical composition, spacing between lines, spacing between characters, number of columns, etc. Based on these input data, the editing machine 12 edits the manuscript. Edit the given desired text. In addition, tape 1
0 and 11 may be various recording means such as floppy disk (trade name) and others.

The text edited by the editing machine 12 is temporarily stored on a magnetic tape 13 as a storage device, and this stored data is calculated together with data in a font memory 14 made up of fonts such as Mincho and Gothic fonts. It is input to the device (CPU) 15.

This arithmetic unit 15 calculates and forms the size, position, etc. of one page of text according to the size, position, and font of the specified input characters, and sequentially stores this in the page memory 16.

Thus, each one stored in the page memory 16
The page worth of text data is given to a laser beam printer 100, where it is printed on plain paper to first create a galley 17 for proofreading. The operator, etc., looks at this galley 17 and checks if there are any mistakes in the size or position of the characters, and if there are any typos or omissions, corrects the tape 10, and if there is a mistake in the size or position of the characters, corrects tape 11 and proofreads the text. In this way, once the printed galley 17 is free of mistakes, a block copy 18 is created next, and this block copy 18 is sent to the printing process.

Here, an example of the configuration of the laser beam printer 100 will be described with reference to FIG. A photosensitive drum 105 charged with
is exposed according to the scanning.

The exposed photosensitive drum 105 passes through the developing section 107 and reaches the position of the transfer charger 108, where it is transferred onto transfer paper (plain paper) 110 fed from the hopper 109, and fixed at the fixing section 111. After the power is stored, the power is stored in the power stacker 112.

Note that the transfer paper 110 is placed at the transfer charger 108 position.
The photosensitive drum 105 after being transferred to the static eliminating section 11
3, the charge is removed by a cleaning unit 114, and the battery is sent to a charging unit 106.

Thus, by controlling the light polarizer 103 and the rotating mirror 104 based on the data from the page memory 16, the text stored in the page memory 16 is transferred onto the transfer paper 110, and this is obtained as a galley 17 or a block copy 18. It will be done.

As described above, according to the present invention, if a printed kanji manuscript, a typeface type manuscript, a typeface clearing plate, etc. is placed in an optical reading area, one character is cut out from a kanji string, recognized by OCR, and then read. record kanji,
By loading this data onto a computer phototypesetting machine, it is possible to output on plain paper, reducing costs, and eliminating the need for the printing process, which improves work efficiency and prevents work errors. .

Although tapes 10 and 11 are used for OCR output and computer phototypesetting input correction, it is also possible to use cards or other means, and they can also be used as storage devices.
It is also possible to use IC memory and other
The circuit configuration of the printed kanji OCR device 9 is not limited to the means shown in FIG.

[Brief explanation of drawings]

FIG. 1 is a block diagram of one embodiment of the present invention, and FIG.
The figure is an explanatory diagram showing the configuration of the laser printer. 1, 2...Manuscript, 3...CRT display and keyboard, 4...Tablet cursor, 5...
Mechanism scanning unit, 6... Preprocessing/recognition unit, 7... Dictionary pattern memory (core memory), 8... Control device, 9... Printed Kanji OCR device, 10, 11
...tape, 12 ... editing machine, 13 ... magnetic tape, 14 ... font memory, 15 ... arithmetic unit (CPU), 16 ... page memory, 17 ... galley,
18... Version copy, 19... Layout manuscript supply system,
100...Laser beam printer.

Claims (1)

[Claims] 1. A CRT display and keyboard for specifying any part of a document as a reading area through dialogue, as well as a tablet and a cursor for inputting coordinates are provided, and the document on the reading paper is vacuum-adsorbed on the reading stand. It is equipped with a mechanical scanning section that sequentially freely conveys the document, scans the surface of the document with a reading head consisting of a self-scanning element of a charge-coupled device, performs photoelectric conversion, and sends out a binary pattern signal a. a is introduced to predict the character position in one line, correct the irregular character arrangement, extract the characters in one line, perform sampling, calculate the complexity index from the detected extracted character pattern, and calculate the complexity index from the sampled pattern. In addition to extracting the four-sided code, a candidate character for the input pattern is selected from among all recognition target characters based on these two parameters, and a candidate character signal b, a dictionary pattern signal c, and a similarity signal d are sent out. It has a recognition unit and receives a candidate character name signal d and a dictionary pattern signal c.
It has a memory that accepts the dictionary pattern and stores the dictionary pattern, and further sends the stored dictionary pattern signal c to the preprocessing/recognition section, and operates each section to determine the reading result from the similarity signal d and output it to the magnetic recording medium. A kanji OCR device that optically reads characters formed by a control device, a device that records kanji data recognized by the kanji OCR device on a magnetic recording medium, and a character code and code that designate the stored predetermined characters. An editing machine that edits text by inputting character size and position data, a storage device that temporarily stores the output of the editing machine, and a storage device that stores print data for each page from the output of the arithmetic device. 1. A printing kanji OCR device for creating printing blocks, characterized by comprising a page memory for outputting the page memory, and a laser beam printer for printing on plain paper by the output of the page memory.