JPS5856076A - Optical character reader - Google Patents

Abstract

PURPOSE:To output a readout result having the same intervals that a text original has, by recognizing and storing characters in plural lines in a line pattern memory, finding a base line position through controlling, and then automatically line intervals while eliminating the need to specify the line intervals. CONSTITUTION:A photoelectric conversion part 1 scans characters on a form and converts their character patterns into binary signals, which are applied to a recognition part 2; and the part 2 recognizes the character patterns to store character codes in a buffer memory 4 temporarily. Further, character patterns for one line of a text form are stored in the line pattern memory 3 of the recognition part 2 and under the control of a control part 6, the character patterns are read, character by character, out of the memory 3 to calculate the similarity with standard patterns selected in a dictionary memory 8 corresponding to the kind of a font, thus recognizing the characters. On this one-line character recognition, the control part 6 finds the center position of the characters stored in the memory 4 to find the base line position of each character pattern on the basis of correction data set up in a working memory 7, thereby outputting a readout result having the same intervals that the text form has.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an optical character reading device for inputting text such as documents and sentences to a word processor.

Generally used for inputting text into word processors.
Optical character IIl! The scanning device must read the line spacing of the text original as it is and input it to the word processor. In the conventional optical character reading device, the operator looks at the line spacing of the text document and specifies the line spacing to be read from the text document using a line spacing designation switch provided on the operating panel. This line spacing specification is complicated, and if the specification is incorrect, the following problems will occur.

For example, if the line spacing is 3 lines/inch and you specify 6 lines/inch, unnecessary blank lines will be inserted into the reading results. Conversely, if the line spacing is 6 lines/inch and specified as 3 lines/inch, missing lines will occur in the reading results. In addition, it is difficult to mix and read text documents with different line spacings at the same time.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an optical character reading device that can eliminate these conventional drawbacks, automatically detect line spacing, and output reading results with the same line spacing as a text original.

Hereinafter, one embodiment of the present invention will be described in detail using the drawings. In Figure 1i, the photoelectric conversion unit (1) optically scans the characters recorded on the paper, photoelectrically converts them, and then converts them into binarized characters, converting the character image on the paper into binarized characters. Output as a pattern. Although not shown, the photoelectric conversion section (1) includes
An optical system consisting of a light source, a mirror, a lens, and a -dimensional solid-state image sensor (linear image sensor) is provided. The surface of the paper is illuminated by a light source, and the character image recorded thereon is focused on the light-receiving surface of the -dimensional solid-state image sensor by the action of the mirror and lens. −
The interior of the dimensional solid-state sensor includes a large number of minute photodiodes arranged in rows. Each photodiode receives reflected light from a corresponding microscopic area on the surface of the paper or the conveyance path of the paper.

Therefore, this minute area becomes the minimum unit of photoelectric conversion. This minute area is called a dot.

The bits that are then binarized are called white bits, and the bits that are binarized after charging and converting dots with low reflectance are called black bits. Therefore, the charge converter (1) converts the character image on the paper into a character pattern that is a set of white pits and black bits and outputs the character pattern. The recognition unit (2) recognizes characters by calculating the similarity between the character pattern output from the photoelectric conversion unit (1) and the standard pattern stored in the dictionary memory @), and converts the result into a character code ( For example, ASCII
Output as I).

A line pattern storage (3) provided in the recognition unit (2) stores a character image obtained by scanning the character image of one line of paper by the charge conversion unit (1).

The buffer memory (4) temporarily stores the character code output from the recognition unit (2). Interface part (5
) has a function of outputting the character code output from the buffer memory (4) to an external device.
This external device refers to an English word processor.

The dictionary memory connected to the recognition unit (2) stores n types of dictionaries (JI) to (JN) corresponding to n types of printing 7 onts. Each dictionary consists of standard patterns for each character in its corresponding printed 7 onts. Control 5 (6) controls the entire device. A working memory (η) connected to the control unit (6) temporarily stores the processing results of the control unit (6).

The operation will be explained below. FIG. 2 shows a text sheet (20) on which text to be read by the optical character reading device of the present invention is recorded. The code of this text paper (20) (
21) to (24) indicate the first to fourth character lines in which text is recorded, respectively. In the photoelectric conversion unit Q), this text paper (20) is scanned in the direction of arrow B while being conveyed in the direction of arrow A. The nine-binarized pattern obtained by the single scan is stored in the row/turn memory (3) in the recognition unit (2).

This row pattern memory (3) is ``&'' so that it can store the binarized pattern obtained by scanning 128d. In this row pattern memory (3), the binarized pattern obtained in the 129th scan is the same as the 2 value pattern obtained in the 1st scan.
Stored after deleting the value pattern. Similarly below,
The binarized patterns obtained in the 130th and subsequent scans are also stored.

Although not shown, the oldest stored binarization pattern among the binarization patterns stored in the current line pattern memory (3) is stored in the recognition unit (ring) on the text paper (2
It has a counter that indicates how many times the binarized pattern was obtained counting from the tip of 0).

Now, write the first character line (21) on the text paper (20).
) is stored in the line pattern memory as shown in FIG.

Recognition unit (2) reads the binary pattern for each character from the ti row pattern memory (3). By calculating the similarity with this read standard pattern,
What letter is V! be recognized.

This opening result is stored in a buffer memory.

When all characters in the character line (21) have been recognized, the control unit (6) determines the baseline position (yl) of the character line (21) as shown in FIG. That is, since the characters in the character line (21) are not necessarily printed straight, the base line position of each character in the character line (21) is first determined. Control part (6)
scans the contents stored in the line pattern memory (3) to find the center position of each character pattern. The base line position of each character pattern is determined by adding the correction data provided for each character previously stored in the working memory (7) K to this center position. FIG. 4 shows the relationship between characters and base line positions. For uppercase letters and numbers, the base line position (30) is a distance d from the center position. The lowercase letters g, j, and y #i of the alphabet, the center position is the base line position (30) - t'16°, code aprost
The base line position (30) of rophe r s J is a distance e from the center position. Therefore, as described above, it is sufficient to store the distance between the center position and the base line position for each character. K like this and the character line (21
) After finding the base line position of each character pattern, find the average value. The obtained average value is stored in the working memory (7) as the baseline position (yl) of the character line (21).

Thereafter, characters in a plurality of lines are similarly recognized, the results are stored in the buffer memory (7)K, and the base line position thereof is determined.

Next, the line pitch between each character line is calculated by dividing the difference between the base line positions by 5. The minimum line pitch among these determined line pitches is set as the standard line pitch for this text paper. Next, divide the line pitch between each character line by this reference line pitch to determine how many blank lines will be inserted between each character line. The control unit (6) stores line feed symbols in the buffer memory (
4) Edit by inserting into the recognition results stored in K. When this editing is completed, the contents stored in the buffer memory (4) are output to the English word processor via the interface section (5).

In this embodiment, an example in which English text is input has been described, but it can also be applied to Japanese text input.

As is clear from the above description, the present invention has the following effects. No need to specify line spacing, Opel
It is very easy to read texts with multiple types of line spacing, and can read texts with different line spacings. An optical character reading device that can output reading results with the same line spacing as text can be provided.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a diagram showing a text sheet, FIG. 3 is a diagram showing rows (binarized stored in turn memory); Figure 4 is a diagram showing the relationship between the center of a character and a base line. (1) ...Photoelectric conversion section (2) ...Recognition section (3) ...Row pattern memory 373- (4) ...Buffer memory (5)・・−・Interface (6) ・・
...Control unit (7) ...Working memory (8) ・
...Dictionary memory (20) ...Text paper (21) to (24) ... - Character line (30) ... Baseline position (7317) Agent patent attorney Kensuke Norichika (7801)
Representative Patent Attorney Sanno - Figure 2 ↑ Figure 3 Figure 4

Claims (1)

[Claims] a photoelectric conversion section that photoelectrically converts a character image recorded on paper, quantizes it as a binary pattern, and outputs it; a means for storing the binary pattern for one line; and a means for storing the binary pattern for one line. means for recognizing, means for storing the recognition result, means for determining the base line position of each row, means for determining the reference line spacing from the determined baseline position of each row, and means for calculating the reference line spacing between the determined base line spacing and the baseline. means for determining the number of line breaks to be inserted between lines from the position, and the la! 1. An optical character reading device comprising means for inserting a predetermined line feed symbol into a text result and editing the recognition result.