JPH0542529Y2 - - Google Patents

Description

[Detailed Description of the Invention] [Technical Field of the Invention] This invention relates to a pattern recognition device.

[Prior art and its problems] Conventionally, such pattern recognition devices
Some methods are known, such as those that store pattern data read by OCR (optical character reader), etc., as they are, and others that perform character recognition on the pattern data read by OCR, etc., and store them in the form of character code data. ing.

However, in those that store pattern data in its original form, for example, 24 x 24 =
A storage capacity of 576 dots, or 32 x 32 = 1024 dots, is required, which requires a huge amount of memory capacity.
When comparing data for each character, using the example above, it is necessary to compare data for 576 dots or 1024 dots for just one character, which makes data match comparison very difficult. .

In addition, although there is no problem in terms of the required storage capacity and data matching comparison when storing the above character code data, it is difficult to perform character recognition from the read pattern data and convert it into the corresponding character code. There were problems in that it took a long time to process and errors in recognition occurred.

[Purpose of the invention] This invention was made in view of the above-mentioned circumstances, and its purpose is to recognize character patterns that are related to each other, such as a postal code, address, and telephone number. It is an object of the present invention to provide a pattern recognition device that can recognize characters quickly for a certain group of character patterns and has fewer errors in character recognition.

[Key points of the invention] In order to achieve the above-mentioned purpose, this invention, for example, if the postal code is 105, the address is Minato-ku, Tokyo.
Phone numbers are 03 (400) ~ 03 (499), 03 (570) ~ 03
(599), etc., where the fields to which each character pattern group belongs are related to each other, the characters in one field can be predicted from the other field, and the character patterns read. The main point is to correct and recognize the characters accurately.

[Configuration of Embodiment] An embodiment of the present invention will be described below in detail with reference to the drawings.

In this embodiment, the postal code, address, and telephone number of a business card are predicted to some extent from the postal code, and characters are recognized more accurately based on the predicted results. In this example,
The pattern information reading section 12 serves as a pattern reading means, the belonging field specifying key 4 serves as a belonging field specifying means, the address/TEL/address correspondence table 24, the address control section 25, the recognition data ROM 21 serve as a character prediction means, and the recognition section 20 correspond to the recognition means, respectively.

<Overall appearance configuration> Figure 1 shows the overall appearance of the pattern recognition device. A slit-shaped insertion opening 1 is provided on the bottom surface of the pattern recognition device, and a business card A is inserted through this insertion opening 1 to read the business card. The characters on A are read and recognized. A display section 2 is provided above the surface of the pattern recognition device, and this display section 2 displays the read character pattern on the business card A, the company name, name, address, and postal code that have been read and recognized and stored. , a list of phone numbers, etc. will be displayed. A key input section 3 is provided below the display section 2 .

The key input section 3 is provided with a field designation key 4, a search key 5, and a cursor key 6. The field designation key 4 includes a name key 7, a company name key 8,
Address key 9, postal code key 10, phone number key 1
1, each of these keys 7 to 11, together with the cursor key 6, specify the field to which the name, company name, address, postal code, and telephone number belong, for each character pattern group read on the business card A. , name, company name, which was read, recognized and memorized,
It instructs to display a list of each address, postal code, and telephone number together with the search key 5. When specifying the above field of affiliation,
Use the cursor keys 6 to move the tip of the display cursor to the upper left corner of the character pattern group displayed on the display section 2, and press each key 7 to 11 of the belonging field designation key 4.
, then move the tip of the display cursor to the lower right corner and turn on each of the keys 7 to 11 of the belonging field designation key 4.

<Overall Circuit Configuration> FIG. 2 shows the overall circuit configuration of the pattern recognition device, in which the character pattern on the set business card A is read by the pattern information reading section 12 and stored in the pattern information storage section 13. The image is displayed on the display unit 2 through the display memory 14. The character pattern data stored in this pattern information storage section 13 is stored in each key 7 to 1 of the belonging field designation key 4.
Name buffer 1 for each field of belonging specified in 1.
5. The information is transferred to a company name buffer 16, an address buffer 17, a postal code buffer 18, and a telephone number buffer 19, and provided to the recognition unit 20. The read/write control for each buffer 15-19 is performed based on the read/write control signal and specified address data from the control section 22 in response to the operation of each key 7-11 of the key input section 3. Also, the pattern information storage section 1
3. Control of reading and writing character pattern data to and from the display memory 14 is performed by the control section 22 through the address control section 30.

In this recognition unit 20, the recognition data ROM 21 is used for the given character pattern data of each field.
Among the sample pattern data for character recognition stored in the sample pattern data for character recognition, the most similar one is selected, and its character code is sent to the control section 22. The read address of this recognition data ROM 21 is given by the address control section 25, and the read address of the address control section 25 is sequentially incremented by the recognition section 20. The character code recognized and discriminated by the recognition unit 20 is also given to the data judgment unit 23 to determine the character code of the postal code, and if it is a postal code, the character code of this postal code is 〒・
It is given to the TEL/address correspondence table 24.

The 〒/TEL/address correspondence table 24 includes the third
As shown in the figure, all addresses and telephone numbers that apply to each postal code are stored in each address using the postal code as the read address. For example, if the postal code is 152, the address is limited to Meguro Ward, Tokyo, Since telephone numbers range from 03 (710) to 03 (724), these addresses and telephone numbers are stored. All predicted addresses and phone numbers corresponding to each postal code read from this 〒/TEL/address correspondence table 24 are given to the address control unit 25,
When the recognition section 20 performs character recognition, sample pattern data corresponding to the expected address or telephone number is read character by character from the recognition data ROM 21 and provided to the recognition section 20.

The recognition unit 20 corrects the character code recognition of the character pattern data of the address or telephone number given by the emotional buffer 17 or the telephone number buffer 19 to be more accurate based on the sample pattern data of the expected address or telephone number given. It is recognized and given to the control section 22.

The character codes of the company name, address, telephone number, and postal code given to the control unit 22 are written into the company name storage unit 27 through the search unit 26 in the order of company names, as shown in FIG. As shown in FIG. 5, the names are written into the name storage section 28 in the order of their names through the search section 26, and the storage address of the company name in the company name storage section 27 corresponding to each name is written.

When searching for a company name or name, the stored character codes are sequentially read out from the company name storage section 27 and name storage section 28, and converted into character data for display by a character generator 29. The converted image is set in the display memory 14 and displayed on the display section 2.

[Operation of the embodiment] Next, the operation of this embodiment will be described.

<Character Recognition Writing Process> Now assume that a business card A as shown in FIG. 1 is inserted through the insertion slot 1. Then, the character pattern of this business card is read by the pattern information reading section 12, written into the pattern information storage section 13, and then sent to the display section 2 through the display memory 14 as shown in FIG.
will be displayed.

In response, move the tip of the cursor displayed on the display unit 2 to the upper left corner of the company name "XX Co., Ltd." using the cursor key 6, turn on the company name key 8, and then Move the company name to the lower right corner as shown in , and turn on the company name key 8. Then,
The character pattern data of the company name within the rectangular frame drawn with this cursor is transferred to the company name buffer 16 and given to the recognition unit 20 one character at a time, and several approximate sample pattern data are collected from the recognition data ROM 21. are sequentially read out and given to the recognition unit 20 for character code recognition.
is written into the company name storage section 27 via the search section 26.

Next, if you specify the name "Jiro Kataoka" in the same way with the cursor and turn on the name key 7, the character code of the name will be recognized in the same way, and the name will be stored in the name storage section 28 via the control section 22 and the search section 26.
will be written to. At this time, the stored address value of "XX Co., Ltd." in the company name storage section 27 is written into the name storage section 28 as well.

Next, when the postal code "152" is specified with the cursor and the postal code key 10 is turned on, the character code of the "152" postal code is recognized in the same way, and the character code is recognized via the control section 22 and the search section 26. It is written in the company name storage section 27, and the expected address corresponding to this postal code "152" is "Meguro-ku, Tokyo..."
and the expected telephone number "〒3 (710)...〇3 (724)..." is 〒・TEL・Address correspondence table 2
4, the sample pattern data of the expected address and expected telephone number is read out from the data ROM for recognition.
21 is read out one character at a time and provided to the recognition unit 20.

In contrast, if you first specify "2-3-4 Haramachi, Meguro-ku, Tokyo" with the cursor and turn on the address key 9, the predicted address "Meguro, Tokyo" is given to the recognition unit 20 as sample pattern data. If the first half of the address is not much different from the sample pattern data, the character code will be determined almost uniquely.

In this way, when character recognition of an address is performed, the recognition speed becomes extremely fast based on the predicted address, and recognition errors are also extremely reduced.

In this case, if two place names are expected from one postal code, for example, the postal code "189" corresponds to "Higashimurayama City, Tokyo" and "Higashiyamato City, Tokyo". , these two types of predicted addresses are given as sample pattern data, and the one with the character pattern that is closer to the other is selected. This is the same even when there are three or more applicable place names. Even in such cases,
There is no difference in the fact that the recognition speed is fast and there are fewer errors.

Next, the phone number “03 (711) 0001”
In the same way, the character code recognition is performed quickly and with few errors based on the predicted phone number "03 (710) to 03 (724)".

The address and telephone number thus recognized are similarly written into the company name storage section 27 via the control section 22 and the search section 26.

<Reading and display processing> By repeating the above processing, in order to search a list of many preset names, company names, etc., it is sufficient to turn on the name key 7 and the search key 5 for names. Then, the character codes of the names are sequentially read out from the name storage section 28 through the search section 26 by the control section 22, and are sequentially displayed on the display section 2 via the character generator 29 and the display memory 14. Then, when the cursor is moved to a desired name and the search key 5 is turned on, the control section 22 reads out the company name address corresponding to this specified name from the name storage section 28, and
The company name, address, telephone number, and postal code are read out from the address in the company name storage section 27 corresponding to the address and displayed on the display section 2.

Furthermore, in order to search a list of company names, the company name key 8 and search key 5 may be turned on. Then, the control unit 22 sequentially reads out the character codes of the company name from the company name storage unit 27 through the search unit 26, and reads them out from the character generator 29 and the display memory 14.
The images are sequentially displayed on the display unit 2 via the . Then, when the cursor is moved to a desired company name and one of the address key 9, postal code key 10, or telephone number key 11 is turned on, the control section 22 selects the specified company name from the company name storage section 27. The address, postal code, or telephone number corresponding to the address is read out and displayed on the display unit 2. When the name key 7 and search key 5 are turned on while specifying this company name, the control section 22 selects the name storage section 28 that stores the same address as the address of the corresponding company name in the company name storage section 27. The name is searched and displayed on the display section 2.

In this way, the information on the preset business card can be retrieved and displayed on the display section 2 through character recognition.

In addition, the character patterns read by the pattern reading means may be symbols, figures, etc. in addition to characters such as numbers, kana, and kanji, and the related character patterns predicted by the character prediction means include addresses based on postal codes,
It may be a postal code or telephone number based on an address rather than a telephone number, or any other character pattern group that has a certain relationship, such as age, occupation, gender, etc., may be predicted. This is not limited to the examples in any way, and the same applies to other means and names.

[Effects of the invention] As explained in detail above, this invention has the following effects: For example, if the postal code is 105, the address is Minato-ku, Tokyo, and the telephone number is 03 (400) - 03 (499), 03 (570) - 03 ( 599), etc., where the fields to which each character pattern group belongs are mutually related.
By predicting the characters of one field from another field and accurately correcting and recognizing the characters based on the read character pattern, we can recognize characters that are related to each other when recognizing character patterns. Regarding the pattern group, effects such as faster character recognition speed and fewer errors in character recognition are achieved.

[Brief explanation of the drawing]

The drawings show an embodiment of the present invention; FIG. 1 is an external view of a pattern recognition device, FIG. 2 is an overall circuit diagram of the pattern recognition device, and FIGS.
TEL/address correspondence table 24, company name storage section 2
7. A diagram showing an example of storage in the name storage section 28. FIG. 6 is a diagram showing an example of display on the display section 2. 2...display section, 4...belonging field designation key, 12
...Pattern information reading unit, 13...Pattern information storage unit, 20...Recognition unit, 21...Recognition data ROM, 22...Control unit, 24...〒・
TEL/address correspondence table, 25... address control section, 27... company name storage section, 28... name storage section.

Claims (1)

[Claims for Utility Model Registration] A pattern reading means for reading the character pattern drawn on the writing surface; and a field specifying means for specifying the field to which the character pattern group belongs for a group of character patterns read by the pattern reading means; , a character prediction means for predicting characters from one field to another field for character pattern groups whose fields are mutually related among each character pattern group specified by this belonging field designation means; A pattern recognition device comprising: recognition means for accurately correcting and recognizing the character pattern of the other field read by the pattern reading means based on the prediction result of the character prediction means.