JPS6326789A - Character recognizing device - Google Patents

Abstract

PURPOSE:To save the memory area of a data processing part of a host computer side by displaying a data read out of a field attribute memory for storing an attribute designated by a keyboard. CONSTITUTION:The character attribute of a low recognition rate is stored in a field attribute memory 7 by designating an attribute code from a keyboard 10. Thereafter, when an input form 1 is inputted to a character recognizing part 2, a character scanned by a character segmenting part 3 is stored in a storage memory 6 as an image data by one character each, and also, sent to a recognizing part 4. A host computer 8 provides an image data area in a data processing part by a portion of a character attribute of a low recognition rate. Among the image data stored in the storage memory 6, the image data of the attribute stored in the field attribute memory 7 is led out, sent and stored in the data processing part 9 through a data output part 5, and displayed on a CRT display 11.

Description

Detailed Description of the Invention (a) Industrial Application Field The present invention relates to a recognition device for various characters written on a slip or the like as an input manuscript. The present invention relates to a character recognition device that specifies image data in advance and displays image data on a screen.

(b) Conventional technology Generally, various characters written on an input form are scanned by an optical reader, and compared with a reference pattern in a character recognition section.
The recognition result is displayed on a display element such as a CRT display or an LCD display, and if the predetermined character is output and correct, it is stored in various memory means. A method is often adopted in which the data is modified and then stored in the memory means.

In this case, the characters written on the input form include characters of a plurality of attributes, such as kanji, kana, numbers, alphabets, and symbols.

The large number of characters written on the input form have attributes that are highly recognized by the recognition unit and those with low attributes.The results recognized by the recognition unit are transmitted to a display device and displayed for the operator (operator) to recognize. ) distinguishes between correct and incorrect characters, and selects the incorrect character from among the correct character candidates for jetting, or sequentially corrects the incorrect character using a host computer unit equipped with a word processing function.
No. 5088.

(c) Problems to be Solved by the Invention In the above-mentioned prior art, when simply sending the image data of a misrecognized character to the host computer, a memory is provided to temporarily store it, and the data is rejected. In the method of storing character image data in the memory, all characters must be stored in the image memory area when reading the input document, which requires a large memory area.

The character recognition device of the present invention provides a novel character recognition device that eliminates the above-mentioned drawbacks.

(d) Means for Solving the Problems The present invention provides an image memory that stores image data obtained by scanning an input form on which characters including kanji, kana, alphabets, numbers, and symbols are written, and A character recognition section that has a character cutting section that cuts out characters, a recognition section that recognizes output data from the character cutting section, a host computer section that has a keyboard, a data processing section, and a display means as an output device. A field attribute memory for storing attributes designated by the keyboard is provided in the character recognition section, and data read from the field attribute memory is displayed on the display means.

(*) Effect With the configuration of the present invention, the locations where numbers, kana, alphabets, or symbols are written on input forms with a specified format, such as slips, can be specified, so they can be recognized among these attributes. Since only the field of character attributes with a low rate is stored as image data in the memory, the data area is sufficient only for the image data.

(f) To explain the present invention according to the drawings, Fig. 1 is a block diagram of the character recognition device of the present invention, Fig. 2 is a screen state diagram of the device, and Fig. 3 is a memory used in the device. Figure 3 shows a state diagram illustrating stored content.

To explain the present invention according to the drawings, (1) is an input paper;
(2) is a character recognition unit having a character extraction unit (3), a recognition unit (4), a data output unit (5), an image memory (6) and a field attribute memory (7); (8> is a data processing unit (
9), keyboard (10), CRT display (11)
), a host computer having data storage means (12) and a printer (14) for printing on output paper (13).

Next, the operation of the character recognition device of the present invention will be explained.
In Figure 1, the attribute with a low recognition rate, that is, the kanji in the normal case, corresponds to this character attribute, but this character attribute can be entered in advance on the keyboard (10
), the attribute code corresponding to the kanji is designated and stored in the field attribute memory (7) via the data processing unit (9).

After that, when the input paper (1>) is input to the character recognition unit (2), the characters scanned by the character cutting unit (3) are stored one by one as image data in the image memory (6), and the recognition unit (4) At this time, for character attributes with a low recognition rate, such as the above-mentioned kanji attributes, when all character extraction in the corresponding field is completed, image data is sent to the data output unit (5).The recognition unit (4) performs recognition according to the field attributes and sends the character code to the data output section (5).

Further, the data output section (5) distinguishes between the character code sent from the recognition section (4) and the image data sent from the character cutout section (3) and outputs it to the host computer (8).
) is sent to the data processing unit (9).

In this case, use the following data structure as an example.

The host computer (8) provides an image data area in the data processing unit (9) in the host computer () for the field of character attributes with poor recognition rate, in the above example, the kanji attribute, and uses it for correction. It will be done. Next, among the information stored in the image memory (6), the attributes specified in the field attribute memory (7>) and other information are divided into three groups, A, B, and C, as shown in Figure 2. However, in the illustrated example, since the second and fifth lines have kanji attributes, it is specified in advance using the keyboard (10) and the character cutout part (3) is inputted from the field attribute memory (7). Image data is accessed in accordance with the designation and derived from the image memory (6), sent to the host computer (8) through the data output section (5), and displayed on the CRT display (
11) It can be displayed on top.

To explain this pattern using Figure 3, the image memory (6
) can store five lines of image data as shown in FIG. 7), the data processing unit (9) in the host computer (8) stores the image data of the two lines corresponding to the kanji attribute as shown in FIG. 3(b), Therefore, only fields for character attributes that are poorly recognized, such as kanji attributes, are required.

Here, the second
A so-called text display is performed on the A part of the diagram, a so-called graphic display is performed on the B part, and a correction mode is applied to the kanji attributes in the B part, that is, the cursor is moved to a predetermined kanji position and the similarity is calculated. The candidate kanji characters are displayed in ascending order and selected by key operation on the keyboard (10) to replace the candidate kanji displayed in the C section on the CRT display (11).

Therefore, in the graphic display area, image data is displayed as the data for the corresponding correction line, so the input form (
1) is not necessary.

When the correction work is completed, the correct reading result is transferred from the RAM, etc. as a temporary memory of the data processing unit to the storage means (12) such as a floppy disk or hard disk.
) 6 Next, when printing on output paper, start printing by operating the keyboard (10), and then the storage means (12)
The reading results of the input paper stored in can be printed on the output paper <13).

(g) Effects of the Invention According to the character recognition device of the present invention, among the characters with various attributes written on the input form, such as kanji, kana, alphabets, or numbers, kanji can be cited as an attribute with a low recognition rate. However, by specifying this kanji attribute in advance from the keyboard, the image data in the image memory is displayed on the display device, and the characters on the input paper are compared with the reading results on the display device, so there is no need to look at them. Characters with attributes with a low recognition rate can be verified on the screen, which not only simplifies the correction work, but also requires less memory area in the data processing section of the host computer, and the memory capacity is small. As a good advantage, the present invention is extremely effective when used in a character recognition device of an optical reader, which is generally called OCR.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the apparatus of the present invention, FIG. 2 is a screen diagram of the apparatus, and FIG. 3 is a diagram of the memory used in the apparatus. (1)...Input paper, (2)...Character recognition unit,
(3)...Character cutting section, (4)...Recognition section,
(5)...Data output section, (6)...Image memory,
(7)...Field attribute memory, (8)...Host computer, (10)...Keyboard, (1
1)...CRT display, (13)...output paper. Applicant Sanyo Electric Co., Ltd. and one other agent Patent attorney Takuji Nishino and one other person Figure 1 Figure 2 Figure iJ3

Claims (1)

[Claims] 1. An image memory that stores image data obtained by scanning input paper on which characters including kanji, kana, alphabets, numbers, and symbols are written; a character cutting section that cuts out characters from the image memory; and a character cutting section that cuts out characters from the image memory. a character recognition unit having a recognition unit that recognizes output data of the character recognition unit, a host computer unit having a keyboard, a data processing unit, and a display means as an output device, and storing attributes specified by the keyboard. A character recognition device characterized in that a field attribute memory is provided in the character recognition section, and data read from the field attribute memory is displayed on the display means.