JPH0632088B2 - Character recognition method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition method for an optical character reading device, and in particular, it is suitable for reading many types of character shapes, for example, type characters printed by many types of printers. It relates to a method of recognizing characters in a font.

[Background of the Invention]

As a method of reading characters in different fonts, for example, Japanese Patent Publication Sho 56
-44455, JP-B-59-3793 and JP-A-58
As described in Japanese Patent No. 51390, multiple dictionaries corresponding to each font are prepared, the read character patterns are recognized using all dictionaries, and the result obtained by the dictionary with the highest recognition rate is read and output. There is a way. This method has the advantage that the number of character shapes can be increased without changing the format of the form, but when the number of character shapes increases, it has the drawbacks that the recognition results cannot be sorted by each dictionary and the reading speed becomes slow.

[Object of the Invention]

An object of the present invention is to provide a method for recognizing characters of different fonts at high speed and with high accuracy.

[Outline of Invention]

A feature of the present invention is that, in a method for recognizing characters of different fonts using a character recognition dictionary composed of a plurality of dictionaries for each font, it consists of a specific symbol or combination of characters assigned to each different font. Prepare a glyph identification dictionary, print or print a specific symbol or character assigned to the font used for the form in a specific area of the form, and recognize the symbol or character in this specific area using the glyph identification dictionary. Then, the dictionary of the font style corresponding to the recognition result is selected from the character recognition dictionary, and the character recognition of the form is performed using the selected dictionary.

Example of Invention

An embodiment of the present invention will be described in detail below with reference to FIGS. 1 to 4.

FIG. 1 is a block diagram including some mechanical parts of an optical character reader to which the present invention is applied, FIG. 2 is a diagram showing a format example of a form, FIG. 3 is a diagram showing an example of a glyph identification dictionary, and FIG. The figure is a diagram showing an example of a character recognition dictionary including a plurality of dictionaries for each font.

Referring to FIG. 1, the form 1 is conveyed on a conveyance path 2 where the control unit 7 controls the paper feed. The characters on the form 1 are irradiated on the light source 3, and the reflected light is input to the photoelectric conversion unit 4 including a CCD sensor or the like. The photoelectric conversion unit 4 converts the input optical signal into an electric signal and outputs it to the binarization circuit 5. The binarization circuit 5 binarizes the input electric signal,
A binary signal is output to the memory 6 under the control of the control unit 7. The memory 6 sequentially stores the input binary signal under the control of the control unit 7. The memory 6 stores, for example, a character pattern for one form.

FIG. 2 shows the format of the form 1. The form has a data field 11 on which data to be originally read is printed or printed, and a symbol or character for identifying the glyph of the used character is printed or printed. A glyph identification field 10 is provided.
Are arranged at the upper right corner of the form 1. When printing or printing on the form 1, a specific symbol or character assigned to each printing device such as a printer to be used is printed or printed on the glyph identification field 10. This particular symbol or character corresponds to the character pattern registered in the glyph identification dictionary 12 as described later.

Next, the character shape identification dictionary and the character recognition dictionary will be described in detail with reference to FIGS. 3 and 4.

FIG. 4 shows a character recognition dictionary 13, which is prepared for each character font, for example, for each printer character shape to be used, as shown in 13 (1) to 13 (n). FIG. 3 shows a character shape identification dictionary 12, which is selected by combining one character from each dictionary 13 (1) to 13 (n) of the character recognition dictionary 13 and combining them. That is, the character type “1” is selected from the dictionary 13 (1), the character type “2” is selected from the dictionary 13 (2), and the character type “0” is selected from the dictionary 13 (n). When selecting, it is desirable to use characters of dissimilar types between the dictionaries.

In the above description, the symbols or characters printed or printed in the glyph identification field 10 on the form 1 are specific symbols or characters assigned to the printing device used, and these are registered in the glyph identification dictionary 12. Although it has been described that it matches the character pattern, if the font of the printer used is a character corresponding to the character recognition dictionary 13 (1), it is registered in the character recognition dictionary 12 in this printer. Dictionary 13
The character "1" selected from (1) is assigned.

Now, let us return to the description of FIG. 1 again. The memory 6 stores the character pattern of the form 1. The character cutout unit 8 receives the information necessary for the cutout from the control unit 7, cuts out the character pattern in the character shape identification field 10 on the memory 6, and sends the position information of the character pattern on the memory 6 to the recognition unit 9. The search for the glyph identification field 10 in the character cutting section 8 can be performed by the control section 7 obtaining the position of the glyph identification field on the form 1, for example, the size from the left end and the upper end of the form 1. The cutting out of the character pattern in the character shape identification field 10 is performed by, for example, a method of recognizing the presence of a character component by tracing the contour of the character pattern and allocating the characters one by one according to the printing pitch of the character. The recognition unit 9 stores the character pattern in the memory 6 based on the position information from the character cutting unit 8.
It is recognized by using the read glyph identification dictionary 12. That is, it is determined which character in the character shape identification dictionary 12 the character pattern is closest to. As a specific method, for example, the center of the character pattern and the standard pattern of the dictionary are overlapped with each other, and the ratio of the number of coincident black bits and the number of black bits of the standard pattern is obtained,
A method in which a character pattern with a high ratio is used as an answer is used. The result of character shape identification is sent to the control unit 7. The control unit 7 then sends the information necessary for cutting to the character cutting unit 8 in order to recognize the character in the data field 11. The character cutout unit 8 cuts out the character pattern in the data field 11 character by character based on the same cutting method as described above. Further, the control unit 7 sends the above-mentioned character shape identification result to the recognition unit 9 as a character shape number. The recognition unit 9 selects a dictionary corresponding to the glyph number received from the control unit 7 from a plurality of dictionaries 13 (1) to 13 (n), and uses the dictionary to perform a recognition method similar to the above in the data field 11 Recognize characters.

According to the present embodiment as described above, before character recognition,
The character pattern printed or printed in the glyph identification field on the form is recognized to determine the glyph, and the dictionary corresponding to this determined glyph is selected to recognize the characters in the data field. Characters can be recognized at high speed. Moreover, the recognition accuracy is high because the recognition is performed using the dictionary that matches the shape of the used character.

In the above embodiment, the control unit, the character cutout unit, and the recognition unit each have an independent structure. However, these use a single microprocessor and control the entire time, character cutout process, and recognition process in time series. May be done. Further, the control unit may be an independent microprocessor, and the character cutout unit and the recognition unit may be one microprocessor to perform the character cutout process and the recognition process in time series. Further, the recognition section may be constituted by an independent microprocessor as the character shape recognition section and the character recognition section, respectively. Further, the character shape identification dictionary and the character recognition dictionary may be stored in one memory, may be an external memory in the recognition unit, or may be built in the recognition unit.

On the other hand, as a control method for printing the symbol or character assigned to each printer in the glyph identification field of the form, it can be performed by the control program of the printer or the host device to which the optical character reader is connected, or the printer itself. It is also possible to have such an operation procedure as a program.

The symbols or characters to be printed or printed in the glyph identification field may have a plurality of digits of the same type of character in order to make the glyph identification more reliable.

The glyph identification dictionary is preferably a combination of types of characters that are similar in distance to each other when recognizing each other, and it is also possible to have multiple types depending on the glyph, and characters other than numbers,
It can also be a symbol.

The glyph identification field can be set to a part of the data field, for example, one digit on the right or one digit on the left, or can be used together with the format identification field of the form.

〔The invention's effect〕

According to the present invention, a character shape of a printed character or a printed character is identified, and a dictionary corresponding to the identified character shape is selected from a plurality of dictionaries for character recognition. And it can be performed with high accuracy. In particular, in the present invention, since a glyph identification dictionary composed of a combination of specific symbols or characters respectively assigned to different fonts is used, this glyph identification dictionary is associated with specific symbols or characters respectively assigned to each font. The specific symbols or characters registered in the character form and registered in the character shape identification dictionary have different character forms. Therefore, when recognizing a character in a specific area, the specific symbol or character recorded in the specific area is matched with the dictionary of the font to be selected by the same character pattern of the same font, and the dictionary not to be selected is selected. On the other hand, since matching is performed by different character patterns of different fonts, the dictionary of fonts to be selected can be selected with high accuracy. Further, there is an advantage that the specific area of the form can be printed by the same printing device used for printing the data of the form.

[Brief description of drawings]

FIG. 1 is a block diagram including a partial mechanical portion showing an embodiment of an optical character reader to which a recognition method according to the present invention is applied,
FIG. 2 is a diagram showing an example of a form, FIG. 3 is a diagram showing an example of the character shape identification dictionary shown in FIG. 1, and FIG. 4 is a diagram showing an example of the character recognition dictionary shown in FIG. Explanation of reference numerals 1 ... form, 2 ... conveyance path 3 ... light source, 4 ... photoelectric conversion unit 5 ... binarization circuit, 6 ... memory 7 ... control unit, 8 ... character cutout unit 9 ... ... Recognition unit, 10 ... Character shape identification field 11 ... Data field, 12 ... Character shape identification dictionary 13 ... Character recognition dictionary

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Masao Dono Inventor Masao Dono 2880, Kozu, Odawara-shi, Kanagawa Hitachi Ltd. Odawara Plant (72) Inventor Satoshi Ito 2880, Kozu, Odawara, Kanagawa Hitachi Ltd. Odawara Inside the factory (72) Inventor Toshihiro Hananoi 2880 Kunizu, Odawara-shi, Kanagawa Stock company Hitachi Ltd. Odawara factory (72) Inventor Kenichi Moto Ken 2880, Kozu, Odawara, Kanagawa Stock company Hitachi Odawara factory (56) References JP-A-58-78277 (JP, A) JP-A-58-78276 (JP, A)

Claims (2)

[Claims] 1. A method of recognizing characters of different fonts using a character recognition dictionary composed of a plurality of dictionaries for each font, and a specific symbol which is the same as the font assigned to each different font but different from each other. Prepare a glyph identification dictionary consisting of combinations of characters, print or print the specific symbol or characters consisting of the same font assigned to the font used for the form in a specific region of the form, and mark the symbol of this specific region. Alternatively, the character recognition is performed using the character shape identification dictionary, the dictionary of the font corresponding to the recognition result is selected from the character recognition dictionary, and the character recognition of the form is performed using the selected dictionary. Character recognition method. 2. The character recognition according to claim 1, wherein the specific symbol or character is selected from each dictionary of the character recognition dictionary and is composed of a different symbol or character for each dictionary. Method.