JP4298904B2 - Character recognition method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a character recognition method used when reading and recognizing characters entered on a predetermined form, and in particular, characters used in an OCR (Optical Character Reader) exchange concentration processing system used in financial institutions. It relates to the recognition method.
[0002]
[Prior art]
In recent years, in order to save labor and rationalize exchange transactions, OCR exchange concentration processing systems have been used in financial institutions and the like. The OCR exchange centralized processing system reads an exchange form with an OCR device installed at a financial institution's sales office, recognizes handwritten characters entered in the exchange form, and then reads the image and character recognition of the exchange form. Host computer that manages the result of the recognition of character recognition results as destination data related to foreign exchange transactions. It is configured to send to.
[0003]
Incidentally, in the OCR exchange concentration processing system as described above, it is necessary to avoid misreading as much as possible for the amount of money which is a priority item in the exchange transaction when performing character recognition. For this reason, usually, on a foreign exchange form, a receptionist at a financial institution that accepts the exchange transaction writes the contents of the entry field adjacent to the entry field where the user who wants the exchange transaction enters the amount. A double column is provided. In other words, with regard to the amount of exchange transactions, character recognition is performed on both the amount entered in the entry column and the amount entered in the double entry column, and the contents entered in the double entry column are given priority by comparing the two. As a result, misreading is reduced to prevent price differences.
[0004]
[Problems to be solved by the invention]
However, in the conventional character recognition method used in the OCR exchange centralized processing system described above, the burden on the operator who corrects unread / misread characters at the centralized center increases for the reasons described below, and the operator However, there is a possibility that the destination data is erroneously transmitted without noticing misread characters.
[0005]
For example, in the entry column on the exchange form, in addition to the number representing the amount by the user, a “¥” mark may be entered at the beginning of the amount. In this case, since the “¥” mark is not a number, it cannot be recognized even if it is read by the OCR device, and is unread. For this reason, the operator of the centralized center, for example, has to perform correction processing for deleting the unread handling recognition result displayed and output by the “?” Mark each time, which is a heavy burden and the correction processing. The speed is also lost.
[0006]
In addition, depending on the politeness of entering characters on the exchange report, there may be differences in the character recognition results between the entry column and the double entry column, but the operator who corrects unread / misread characters in these differences. If you are not aware of this, even if the character recognition result for the entry field is correct, the character recognition result for the double entry field will be given priority, and as a result, incorrect destination data may be transmitted to the host computer. is there.
[0007]
In addition, the amount is entered in numbers in both the entry field and the double entry field on the exchange form, but misreading is likely to occur if the numbers are not politely entered. In particular, for example, “0” and “6”, “1” and “7”, “3” and “8” are easily misrecognized. However, if numbers are entered in both the entry field and the double entry field, the character shapes entered on both sides are similar, and each may be misread as well. In this case, if the operator who corrects the unread / misread character is not aware of the misreading, erroneous destination data is transmitted to the host computer.
[0008]
In particular, in the OCR centralized exchange processing system, “amount” is the most important item because it is used for exchange business, and if the “amount” is processed incorrectly, a serious exchange accident will be caused. Become.
[0009]
Therefore, the present invention can reduce the burden on an operator who corrects misreading and misreading, and can prevent the operator from sending erroneous data in advance, thereby enabling a highly safe system operation. An object of the present invention is to provide a character recognition method.
[0014]
The present invention is a character recognition method devised to achieve the above object. In other words, a service user fills in the entry field and the double entry field from a form having a entry field where the user fills in characters and a receptionist who accepts the use of the service user doubles the contents of the entry field. Character recognition by reading a handwritten character with an OCR device and performing character recognition, and displaying the character recognition result and the image data based on the character recognition result on the terminal device to correct the character recognition result. In the method, the double entry field has a relationship that is uniquely determined from the characters entered in the entry field, and characters that are expressed in a form different from that character are entered, and when character recognition is performed for the double entry field, The character recognition result is replaced with the character form entered in the entry field based on the relationship.
[0015]
According to the character recognition method in the above procedure, even if a number is entered in the entry field, the double entry field is uniquely identified from the original number, for example, kana characters representing kanji numbers and the reading of numbers. Characters that have a fixed relationship and are expressed in a form different from the characters are entered. In the subsequent correction processing, the character recognition for the double entry column is replaced with the same form as the characters entered in the entry column. Therefore, in the entry field and the double entry field, the same contents are represented by different forms of characters, so that it is possible to avoid misreading both in the same manner.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, a character recognition method according to the present invention will be described with reference to the drawings. However, here, a case where the present invention is applied to an OCR exchange concentration processing system, that is, a character recognition method used in the OCR exchange concentration processing system will be described as an example.
[0017]
First, an OCR exchange concentration processing system to which the present invention is applied will be described. FIG. 1 is a block diagram showing an outline of an example of an OCR exchange concentration processing system, and FIG. 2 is an explanatory view showing a display example of a character recognition result.
[0018]
As shown in FIG. 1, the OCR exchange central processing system is connected to a central center 20 connected to an OCR 11 and an OCR client 12 installed in a sales office 10 of a financial institution through a circuit network such as the INS network. The installed file server 21, the correction client 22, the verification client 23, and a communication gateway device (hereinafter abbreviated as “communication GW”) 24.
[0019]
In the centralized OCR exchange processing system configured as described above, after a bank employee of the sales office 10 fills in the necessary items (for example, entering the amount in the double entry column) on the exchange report such as the transfer request form 13 received from the customer, When the transfer request form 13 or the like is set in the OCR 11 (see (1) in the figure), the OCR 11 reads the handwritten characters entered in the transfer request form 13 or the like while following the read operation control from the OCR client 12, and Cutout processing and character recognition processing are performed. Then, the OCR client 12 transfers the read image and the character recognition result obtained by the OCR 11 to the centralized center 20 through a line network such as an INS network (see (2) in the figure).
[0020]
When the read image and the character recognition result are transferred from the OCR client 12, in the centralized center 20, the file server 21 stores the read image and the character recognition result in the database 21a in the file server 21 and temporarily holds them. Keep it. Thereafter, the read image and the character recognition result in the file server 21 are distributed to the correction client 22 operated by the operator of the centralized center 20 (see (3) in the figure).
[0021]
As a result of this distribution, the correction client 22 displays and outputs both the read image and the character recognition result in a state where they are arranged vertically as shown in FIG. However, for example, the character recognition result for the amount of money is displayed in a double column in addition to the entry column on the exchange form, so that the respective read images are displayed side by side, and further below either one ( For example, only the character recognition result in the double column) is preferentially displayed and output. By such a display output, an operator who operates the correction client 22 can visually compare and confirm the read image and the character recognition result, and can correct even if there is an unread or misread character in the character recognition result. become.
[0022]
When the correction of the unread or misread characters in the correction client 22 is completed, the corrected character recognition result and the read image that is the basis thereof are subsequently sent to the verification client 23 operated by the officer of the centralized center 20. (Refer to (4) in FIG. 1). In the verification client 23, as in the case of the correction client 22, the read image and the character recognition result are displayed and output (see FIG. 2), and the character recognition result by the officer is verified.
[0023]
Thereafter, when the verification by the officer is completed and the officer presses the transmission key of the verification client 23, the character recognition result that has been verified is sent to the host computer 30 that manages it as destination data related to the exchange transaction. It is transmitted via the communication GW 24 (see (5) in FIG. 1). By performing the series of processing operations as described above, the OCR exchange centralized processing system completes one exchange transaction.
[0024]
Note that the OCR exchange concentration processing system may use a so-called FAX-OCR technology. That is, as indicated by the wavy line in FIG. 1, if the centralized center 20 includes a FAX server 25, a CRU (Character Recognition Unit) 26, and a CRU server 27, the OCR 11 and Even if the OCR client 12 is not installed, an image obtained from the exchange report such as the transfer request form 13 is transmitted as FAX data to the centralized center 20 through the line network. Correction / verification processing can be performed.
[0025]
Next, a characteristic processing operation in the OCR exchange concentrated processing system configured as described above, that is, a character recognition method according to the present invention will be described.
[0026]
[First processing operation example]
First, an example of a character recognition method will be described. FIG. 3 is an explanatory diagram showing an outline of a first processing operation example as an example of a character recognition method .
[0027]
When the OCR exchange centralized processing system having the above-described configuration is used, when a bank employee of the sales office 10 (hereinafter referred to as “teller”) receives the exchange form such as the transfer request form 13 entered by the customer, the teller Check if everything is filled in. If all necessary items are entered, Teller enters the same number as the amount entered in the entry column by the customer in the double entry column on the exchange report.
[0028]
At this time, as shown in FIG. 3A, the customer may enter a “¥” mark at the head of the amount in the entry field, as shown in FIG. In this case, the teller enters “-(hyphen)” at the beginning of the amount corresponding to the “¥” mark, that is, the amount in the double entry column, when entering the double entry column.
[0029]
This “-” is a specific symbol predetermined in the operation of the OCR centralized exchange processing system, and the character in the entry column corresponding to that symbol is handled as if there was no entry in the entry column. is there. In other words, in the OCR 11, if “-” is entered in the double entry column, any character or symbol entered in the corresponding location in the entry column will be erased for that character or symbol. Do it and treat it as if there was no entry in the entry field.
[0030]
Therefore, even if an unrecognizable symbol such as the “¥” mark is entered in the entry field on the exchange document subject to character recognition, if “-” is entered in the corresponding part in the double entry field, As shown in FIG. 3B, the correction client 22 displays both the “¥” mark and the “−” for the read image, but displays only the number for the character recognition result.
[0031]
Therefore, even if an unrecognizable symbol such as a “¥” mark is entered in the entry field on the exchange report, the operator of the correction client 22 displays, for example, a “?” Mark from the result of character recognition on the exchange report. It is not necessary to delete the unrecognized unrecognized symbols that are output. That is, the burden on the operator of the correction client 22 that corrects unread / misread can be reduced by entering "-" in the double entry column. Along with this, the speed of the correction process can be ensured. This can be said to be particularly effective in an OCR exchange concentration processing system configured to accept character recognition results from a plurality of sales offices 10 by one centralized center 20.
[0032]
Here, the case where the specific symbol entered in the double entry column on the exchange form is “−” has been described as an example, but the present invention is not limited to this and is predetermined. Other symbols may be used as long as they are the same.
[0033]
[Second processing operation example]
Next, another example of the character recognition method will be described. FIG. 4 is an explanatory diagram showing an outline of a second processing operation example which is another example of the character recognition method .
[0034]
In the correction client 22, for example, the same amount of money is entered in the entry field and the duplicate entry field in duplicate, so when the character recognition result is displayed and output, the character recognition results for both are compared, Either one is preferentially displayed and output.
[0035]
Specifically, for example, as shown in FIG. 4, the correction client 22 uses a double-written character except when there is an unread character (indicated by “?” In the figure) only in the character recognition result from the double-written column. Display and output the character recognition result from the field with priority. However, at this time, when the correction client 22 displays and outputs either of the character recognition results, the correction client 22 changes the display color depending on whether or not there is a difference between the two character recognition results. For example, if both character recognition results match, the character recognition result from the double entry column (or entry column) is displayed in blue, but if both character recognition results are different, there is no difference. The display is output in pink instead of blue as in the case (when they match).
[0036]
Therefore, if the operator of the correction client 22 refers to the display color of the result of character recognition, even if there is a difference in the result of character recognition between the entry field and the double entry field, it can be easily noticed. In other words, if the recognition result of the customer entry is different from the recognition result of the operator's double entry, and there is a very high possibility that unread or misread has occurred, attention should be paid to the operator of the correction client 22. It can be inspired.
[0037]
Thereby, in the OCR exchange concentrated processing system, the correction processing by the correction client 22 is surely performed, so that it is possible to prevent the communication GW 24 from sending wrong destination data to the host computer 30. . That is, a highly secure system can be constructed.
[0038]
In this example, when the result of character recognition is the same in the entry field and the double entry field, the output is displayed in blue. The invention is not limited to this, and any color may be used as long as each display color is different.
[0039]
[Third processing operation example]
Next, the character recognition method according to the present invention will be described. FIG. 5 is an explanatory diagram showing an outline of a third processing operation example as an example of the character recognition method according to the present invention.
[0040]
When the sales office 10 accepts the exchange form such as the transfer request form 13 entered by the customer, the teller enters the amount entered by the customer in the entry field in the double entry field on the exchange form. However, at this time, the teller enters the amount of money to be entered in the double column, not a number, but a character that has a unique relationship with the number and is expressed in a form different from the number.
[0041]
Specifically, for example, as shown in FIG. 5 (a), if the number entered in the entry field is “1”, the corresponding entry in the double entry field corresponds to “K” of the katakana. ". If the number entered in the entry column is “2”, the katakana “F” is entered in the double entry column accordingly.
[0042]
Then, the OCR 11 performs reading and character recognition for both the numbers entered in the entry column and the katakana entered in the double entry column. However, at this time, when the OCR client 12 performs character recognition for katakana, the recognition result in the double entry column is replaced with a number based on the relationship between the number in the entry column and the katakana in the double entry column.
[0043]
Specifically, for example, as shown in FIG. 5B, when the katakana “hi” is recognized from the double entry column, the recognition result is replaced with the numeral “1”. If the katakana character “F” is recognized, the recognition result is replaced with the number “2”. Thus, for example, even if the character recognition result is Katakana “Itotototo”, the recognition result is replaced with the number “50000” and is displayed and output by the correction client 22. However, also in this case, the correction client 22 may display and output one of the character recognition results with priority as described above.
[0044]
In this way, if the characters to be entered in the double entry column on the exchange form are not numbers but katakana characters, even if the characters to be entered in the double entry column are not polite, the amount will be misread. Can be prevented. Because, if you enter a number in the double entry field as well as the entry field, if the entry is not polite, the character shape entered on both sides is similar, so each will be misread as well, This is because the operator of the correction client 22 is likely not to notice the misreading, but such a situation can be avoided if the double entry column is filled in with katakana.
[0045]
Accordingly, in the OCR exchange concentrated processing system, it is possible to avoid misreading the amount in advance, and thus it is possible to prevent the communication GW 24 from sending wrong destination data to the host computer 30. That is, a highly secure system can be constructed.
[0046]
Here, the case where katakana is entered in the double entry column has been described as an example, but the present invention is not limited to this, and has a relationship that is uniquely determined with respect to numbers, and the numbers Other characters (for example, Chinese numerals) may be entered as long as the characters are expressed in a different form.
[0047]
The first to third processing operation examples described above may be implemented alone or in combination with any plurality. For example, it can be considered that the first processing operation example and the second processing operation example are combined, or the second processing operation example and the third processing operation example are combined.
[0048]
Further, in this embodiment, the case where the present invention is applied to an OCR exchange concentration processing system used in a financial institution has been described as an example. However, the present invention is an OCR concentration processing used for slip processing in, for example, a distribution business. The present invention is applicable to all systems that perform centralized processing using OCR, such as a system.
[0049]
【The invention's effect】
As described above, according to the character recognition method of the present invention, handwritten characters entered on a form are recognized by the OCR device, and the result of character recognition is displayed on the terminal device and corrected. Even in this case, it is possible to reduce the burden on the operator who performs the correction and to prevent the operator from sending erroneous data. The system operation with excellent safety can be realized without being invited.
[Brief description of the drawings]
FIG. 1 is a block diagram showing an outline of an example of an OCR exchange concentration processing system to which a character recognition method according to the present invention is applied.
FIG. 2 is an explanatory diagram showing a display example of a character recognition result in the OCR exchange concentrated processing system of FIG. 1;
FIGS. 3A and 3B are explanatory views showing an outline of a first processing operation example as an example of a character recognition method, FIG. 3A is a diagram showing an example of entry on a form, and FIG. 3B is a display output of a character recognition result; It is a figure which shows an example.
FIG. 4 is an explanatory diagram showing an outline of a second processing operation example that is another example of a character recognition method ;
FIGS. 5A and 5B are explanatory diagrams showing an outline of a third processing operation example as an example of a character recognition method according to the present invention, FIG. 5A is a diagram showing an example of entry on a form, and FIG. It is a figure which shows the example of a display output of a result.

Claims (1)

Fill out the entry field and the double entry field from a form that has the entry field where the service user fills in the letters and the receptionist who accepted the use of the service user doubles the contents of the entry field. The handwritten character is read by the OCR device to perform character recognition, and the character recognition result and the image data based on the character recognition can be displayed on the terminal device to correct the character recognition result. In the character recognition method,
In the double entry column, there is a relationship that is uniquely determined from the characters entered in the entry column, and characters written in a form different from the characters are entered,
When character recognition is performed on the double entry column, the character recognition result is replaced with the form of characters entered in the entry column based on the relationship.

Images