JP6075158B2 - Character input support program and character input support device - Google Patents

Description

  The present invention relates to a character input support program and a character input support device.

  In recent years, techniques for automatically recognizing characters included in images have been developed. In such character recognition processing, for example, a document including characters such as a slip is optically read by a scanner or the like, and a character is recognized by collating an image of the read document with a stored pattern.

  In addition, as a technique related to character recognition, for example, a combination of the presence or absence of correction included in the first corrected character data by automatic character recognition and the presence or absence of correction included in the second corrected character data by automatic character recognition, A technique for determining the reliability level of the system has been proposed.

JP 2006-99213 A

  By the way, the character string recognized from the document is compared with the character string input while looking at the document by the operator. It becomes possible to do. However, such input support can only notify the operator of the position where the characters do not match.

  In one aspect, an object of the present invention is to provide a character input support program and a character input support device with improved operator convenience.

  In one proposal, a character input support program is provided. The computer executes the following processing according to the character input support program. The computer includes a first character string recognized from the image, an accuracy indicating the accuracy of recognition for each character included in the first character string, and a second character string input according to the input operation. To get. Further, each character included in the first character string and each character included in the second character string are compared on a one-to-one basis from the top, the compared characters are different, and the corresponding accuracy is the first threshold value. The first character as described above is extracted from the first character string. Furthermore, a second character that is adjacent to the first character and has a corresponding accuracy less than the second threshold is extracted from the first character string. And the character pair of the 2nd character string corresponding to a 1st character and a 2nd character is determined as an incorrect input.

  Further, in one proposal, a character input support device is provided that executes processing similar to the processing realized by the character input support program.

  According to one aspect, the convenience for the operator is improved.

It is a figure which shows the structural example and processing example of the character input assistance apparatus of 1st Embodiment. It is a figure which shows the structural example of the slip input system of 2nd Embodiment. It is a figure which shows the hardware structural example of a slip input device. It is a figure which shows the example of an input screen. It is a figure which shows the example of a correction candidate display. It is a block diagram which shows the function example of a slip input system. It is a figure which shows the example of a recognition rate table. It is a figure which shows the example of authentication character data and input character data. It is a figure which shows the example of a recognition rate calculation table. It is a figure which shows the example in which one correction candidate is determined. It is a figure which shows the example in which two correction candidates are determined. It is a figure which shows the example in which a correction candidate is not determined. It is a flowchart which shows the example of the process which supports character input. It is a flowchart (continuation) which shows the example of the process which supports character input. It is a flowchart which shows the example of the process which determines a correction candidate.

Hereinafter, the present embodiment will be described with reference to the drawings.
[First Embodiment]
FIG. 1 is a diagram illustrating a configuration example and a processing example of the character input support device according to the first embodiment.

The character input support device 10 is a device that supports a user's character input operation. The character input support device 10 includes an acquisition unit 11 and a determination unit 12.
The acquisition unit 11 is input in accordance with the character string 2 (first character string) recognized from the image 1, the accuracy indicating the recognition accuracy of each character included in the character string 2, and the input operation. Character string 3 (second character string) is acquired. The image 1 is obtained by reading a document with a device such as a scanner, for example. The character string 3 is input by an operator's operation while viewing this document, for example.

  The determination unit 12 compares each character included in the character string 2 with each character included in the character string 3 on a one-to-one basis from the top. Then, the determination unit 12 extracts, from the character string 2, the first character whose compared characters are different and the corresponding accuracy is equal to or higher than the first threshold value.

  Further, the determination unit 12 extracts, from the character string 2, a second character that is adjacent to the extracted first character and has a corresponding accuracy that is less than the second threshold. Note that the first threshold value and the second threshold value may be the same value.

Then, the determination unit 12 determines that the character pair of the character string 3 corresponding to each of the first character and the second character is an erroneous input.
Here, as an example, the image 1 includes the character string “123”, “123” is recognized as the character string 2 by character recognition, and “132” is input as the character string 3 according to the input operation. Suppose. Further, the first threshold value and the second threshold value are set to “80”.

  In this case, first, the acquisition unit 11 acquires “123” as the character string 2 and “132” as the character string 3. Further, “85”, “90”, and “50” are acquired from the top of the character string 2 as the accuracy.

  The determination unit 12 compares each character of the character string 2 and the character string 3. Then, “2” and “3” are extracted from the character string 2 as characters that do not match the character string 3. The accuracy corresponding to the extracted character “2” is “90”, and the accuracy corresponding to the extracted character “3” is “50”. Therefore, the determination unit 12 extracts “2” whose corresponding accuracy is greater than or equal to the threshold value from the character string 2 as the first character.

  Next, the determination unit 12 extracts from the character string 2 a second character that is adjacent to the first character “2” and whose corresponding accuracy is less than the threshold. The characters adjacent to the first character “2” are “1” and “3”. In the character string 2, the accuracy corresponding to the character “1” is “85”, and the accuracy corresponding to the character “3” is “50”. Therefore, “3” having a corresponding accuracy less than the threshold is extracted from the character string 2 as the second character.

  The determination unit 12 specifies a character pair of the character string 3 that respectively corresponds to the first character “2” and the second character “3” extracted from the character string 2. Of the characters in the character string 3, the character at the same position as the first character “2” is “3”, and the character at the same position as the second character “3” is “2”. For this reason, “32” is specified as the character pair. The determination unit 12 determines that an erroneous input has occurred in the character pair “32” in the character string 3. Specifically, it is determined that the character pair “32” is originally “23” but is input with the order changed.

  Here, as an erroneous input of an operator who is accustomed to the key input operation, such as an operator at a store window, there is a high probability of an erroneous input in which the input order is changed. If the input order is incorrect, the input is incorrect for two adjacent character pairs in the input character string.

  Regarding the position of the second character “3” extracted by the above processing, because the recognition accuracy is less than a predetermined threshold value, it has been erroneously input, has been erroneously recognized in character recognition, or is correctly input. It is difficult to judge whether it was done. However, the position of the adjacent first character “2” may be erroneously input because the recognition accuracy is equal to or higher than a predetermined threshold value and the characters do not match between the character strings 2 and 3. high. As described above, in view of the high probability of erroneous input that the input order is switched, in the character string 3, the character “3” corresponding to the first character “2” having the high probability of erroneous input; The character “2” corresponding to the second character “3” that is adjacent to the first character “2” and in a state where accurate determination cannot be performed is considered to be highly likely to have been input after being replaced. be able to.

  On the other hand, in the character string 2, for the character “1” adjacent to the first character “2”, the recognition accuracy is equal to or higher than a predetermined threshold value, and the character comparison results match. From this, it is considered that there is a high possibility that the character “1” of the character string 3 corresponding to the character “1” of the character string 2 is correctly input.

  For the above reason, the determination unit 12 determines that the character pair “32” corresponding to the first character “2” and the second character “3” among the characters of the character string 3 is an erroneous input. As a result, the character input support device 10 can determine an erroneous input in consideration of a statistical erroneous input occurrence status, not simply based on a match determination of each character of the character strings 2 and 3. Become. Therefore, convenience for the operator can be enhanced.

  For example, even if the second character matches the corresponding character in the character string 3, the operator corresponds not only to the character in the character string 3 corresponding to the first character but also to the second character. It can be determined that the character of the character string 3 is also an erroneous input. Therefore, the operator can immediately recognize an erroneous input that cannot be recognized only by the matching determination of each character in the character strings 2 and 3.

  Further, as described above, since the accuracy of recognition of the second character “3” is less than a predetermined threshold, it is difficult to accurately determine an erroneous input or the like. However, even for such characters, it is possible to determine whether or not an erroneous input has occurred by the above processing based on the new determination criterion using the matching determination result and the recognition accuracy for adjacent characters.

Then, the determination unit 12 determines that a pair of “3” corresponding to the second character “2” and “2” corresponding to the third character “3” in the character string 3 is a candidate for erroneous input. .
Here, the character string recognized from the document is compared with the character string entered while looking at the document by the operator. It was only possible to notify the operator of the mismatched position.

According to the above processing, the second character string pair corresponding to the second character and the third character can be determined as an erroneous input candidate. Therefore, the convenience of the operator who performs the input operation is improved.
[Second Embodiment]
Next, an example will be described in which the character input support device according to the first embodiment is applied to a slip input system in which an operator inputs characters while looking at a slip entered by a customer. Similar to the character input support device of the first embodiment, the slip input system determines an erroneous input candidate and presents correction candidates to the operator for the pair determined as an erroneous input candidate.

  FIG. 2 is a diagram illustrating a configuration example of a slip input system according to the second embodiment. The slip input system 5 includes a plurality of scanners 20, a plurality of slip input devices 100, and a recognition rate DB server 200. One scanner 20 is connected to each slip input device 100. The recognition rate DB server 200 is connected to each slip input device 100 via the network 30.

  The scanner 20 optically reads a slip handwritten by a customer and generates image data. The scanner 20 has an OCR (Optical Character Reader) function. That is, the scanner 20 recognizes characters included in the image by collating the read image with a previously stored pattern. Hereinafter, such character recognition is referred to as “OCR recognition”, and a character string formed by OCR recognized characters is referred to as a recognized character string. The scanner 20 converts the generated image data into document data including character data recognized by OCR. The document data includes, for example, a character code of a recognized character string for each input item. In addition, the scanner 20 transmits the converted document data to the slip input device 100.

  The recognition rate DB server 200 is a DB server that stores information regarding the recognition rate for each type of character that can be OCR recognized. The recognition rate is a numerical value indicating the accuracy of character recognition by OCR recognition. In the present embodiment, the recognition rate is used as a parameter for determining the accuracy indicating the certainty of recognition for each character that has been OCR recognized.

  The recognition rate DB server 200 periodically receives information (for example, the number of times of OCR recognition, the number of correct answers of OCR recognition, etc.) for calculating the recognition rate for each character type from each slip input device 100 (for example, sales). At the end). The recognition rate DB server 200 calculates a recognition rate for each character type based on the result of collecting information for calculating the recognition rate. The recognition rate DB server 200 transmits information indicating the calculated recognition rate to the slip input device 100 periodically (for example, at the start of every business).

  A plurality of recognition rate DB servers 200 including a backup DB server may be provided. In this case, control is performed so that the information regarding the recognition rate of each character type matches.

  The slip input device 100 is a computer for an operator to input characters entered on a slip. In addition, the slip input device 100 has a function of assisting an operator in inputting characters.

  The slip input device 100 generates an input screen for an operator to input characters entered in the slip based on the document data received from the scanner 20, and displays the generated input screen on a display. The slip input device 100 acquires a character string input on the input screen by the operator as input character data.

  Further, the slip input device 100 receives information indicating the recognition rate from the recognition rate DB server 200 for each type of character that can be OCR recognized. In addition, the slip input apparatus 100 generates recognized character data by associating each character of the recognized character string by OCR recognition with the received recognition rate.

  Further, the slip input device 100 compares the acquired input character data and the generated recognized character data for each input item, and specifies an input item that may be erroneously input by the operator. The slip input device 100 displays the specified input item with, for example, a color changed. In this case, the slip input apparatus 100 may emphasize the character string input to the specified input item by changing the type or size of the font.

  The slip input apparatus 100 determines a correction candidate for an input item that may have been erroneously input, based on the acquired input character data and the generated recognized character data. The slip input apparatus 100 presents the determined correction candidates on the display. Thereafter, when the operator selects a correction candidate, the input character data corresponding to the input item that may have been erroneously input is corrected.

  The slip input apparatus 100 aggregates information for calculating the recognition rate for each character type. In addition, the slip input device 100 transmits the collected information to the recognition rate DB server 200 periodically (for example, at the end of every business).

  Note that the OCR recognition function may be included in the slip input device 100 instead of the scanner 20. In this case, the scanner 20 transmits the image data of the slip entered by the customer to the slip input device 100. Then, the slip input device 100 generates recognized character data based on the received image data.

  FIG. 3 is a diagram illustrating a hardware configuration example of the slip input apparatus. The slip input device 100 includes a processor 101, a RAM (Random Access Memory) 102, an HDD (Hard Disk Drive) 103, an image signal processing unit 104, an input signal processing unit 105, a disk drive 106, a communication interface 107, and a device connection unit 108. Have. The unit is connected to the bus 109 in the slip input device 100. The recognition rate DB server 200 can also be realized using the same hardware as the slip input device 100.

  The processor 101 is a processor including an arithmetic unit that executes program instructions. The processor 101 loads at least a part of the program and data stored in the HDD 103 into the RAM 102 and executes the program. The processor 101 may include a plurality of processor cores. The slip input device 100 may include a plurality of processors. The slip input apparatus 100 may perform parallel processing using a plurality of processors or a plurality of processor cores. In addition, a set of two or more processors, a dedicated circuit such as a field-programmable gate array (FPGA) or an application specific integrated circuit (ASIC), a set of two or more dedicated circuits, or a combination of a processor and a dedicated circuit is referred to as a “processor”. You may call it.

  The RAM 102 is a volatile memory that temporarily stores programs executed by the processor 101 and data referred to by the programs. Note that the slip input device 100 may include a type of memory other than the RAM, or may include a plurality of volatile memories.

  The HDD 103 is a non-volatile storage device that stores software programs and data such as an OS (Operating System), firmware, and application software. Note that the slip input device 100 may include other types of storage devices such as a flash memory, and may include a plurality of nonvolatile storage devices.

  The image signal processing unit 104 outputs an image to the display 21 connected to the slip input device 100 in accordance with an instruction from the processor 101. As the display 21, a CRT (Cathode Ray Tube) display, a liquid crystal display, or the like can be used.

  The input signal processing unit 105 acquires an input signal from the input device 22 connected to the slip input device 100 and notifies the processor 101 of the input signal. As the input device 22, a pointing device such as a mouse or a touch panel, a keyboard, a numeric keypad, or the like can be used.

  The disk drive 106 is a drive device that reads programs and data recorded on the recording medium 23. Examples of the recording medium 23 include a magnetic disk such as a flexible disk (FD) and an HDD, an optical disk such as a CD (Compact Disc) and a DVD (Digital Versatile Disc), and a magneto-optical disk (MO). Can be used. The disk drive 106 stores the program and data read from the recording medium 23 in the RAM 102 or the HDD 103 in accordance with an instruction from the processor 101.

The communication interface 107 communicates with other information processing apparatuses (for example, the recognition rate DB server 200) via a network such as the network 30.
The device connection unit 108 is a communication interface for connecting a peripheral device (for example, the scanner 20) to the slip input device 100.

  The slip input device 100 may not include the disk drive 106, and may not include the image signal processing unit 104 or the input signal processing unit 105 when accessed exclusively from another information processing device. . The display 21 and the input device 22 may be formed integrally with the case of the slip input device 100.

The recognition rate DB server 200 can also be realized using the same hardware as the slip input device 100.
FIG. 4 is a diagram illustrating an example of the input screen. The input screen 21 a is an input screen displayed on the display 21 by the process of the slip input device 100. The input screen 21a has a plurality of input items and a key assignment display 21b. The input item is an item to be input by the operator. For example, the input screen 21a includes input items for a bank name, a branch name, a subject (account number), an amount, a payee, and a client. Each input item has an input field and an input image.

  The input screen 21a has an input field for each input item. The input column is an area where characters input by the operator are displayed. For example, when the operator selects an input item, a cursor is displayed in the input field of the selected input item. When the operator inputs characters in this state, the input characters are displayed in the input field. On the input screen 21a of FIG. 4, a cursor is displayed in the amount input field, and the user can input characters in this input field.

  A character string that has been OCR recognized by the scanner 20 is displayed above each input field. The display of the character string is for enabling the operator to compare the character string recognized by OCR and the input character string, for example.

The key assignment display 21b displays input characters assigned to the numeric keypad. The character assigned to the numeric keypad varies depending on, for example, the input field where the cursor is displayed.
FIG. 5 is a diagram illustrating an example of correction candidate display. The correction screen 21c is a screen that displays correction candidates for input items that may have been erroneously input. The correction screen 21c is displayed, for example, over the input screen 21a in FIG. As will be described later, the correction screen 21c displays a pair of characters that may have been input in a reversed order. The correction screen 21c has an erroneous input item 21c-1 and an operation item 21c-2.

  In the erroneous input item 21c-1, input items that may have been erroneously input are displayed. The erroneous input item 21c-1 has an input field. A character string that may have been input incorrectly is displayed in the input field. The input field is in a state where input is impossible until re-input is selected from the operation item 21c-2. In addition, a character string that has been OCR recognized by the scanner 20 is displayed above the input field.

  The operation item 21c-2 includes items “candidate # 1”, “candidate # 2”, “re-input”, and “next”. In the “candidate # 1” and “candidate # 2” items, correction candidates in the input field of the erroneous input item 21c-1 are displayed. There are one or two correction candidates for one input item. When there is one correction candidate, only “candidate # 1” is displayed, and when there are two correction candidates, “candidate # 1” and “candidate # 2” are displayed.

  In the example of FIG. 5, the character string “10” is displayed as a candidate for “candidate # 1”. This indicates a candidate for correcting the character string “01” in the sixth and seventh digits from the top of the input character string to the character string “10” by replacing the front and rear. In “candidate # 2”, the character string “51” is displayed as a correction candidate. This indicates a candidate for correcting the character string “15” in the 7th and 8th digits from the top of the input character string to the character string “51” by replacing the front and back.

  The item “re-input” indicates that the character string of the erroneous input item 21c-1 is re-input from the first character. By selecting “re-input”, the input field of the erroneous input item 21c-1 can be input.

The item “next” indicates that the correction of the character in the erroneous input item 21c-1 is confirmed and the next erroneous input item is displayed.
A corresponding key is displayed at the left end of each selection item. For example, in the correction screen 21c, when the operator presses the key “1”, “candidate # 1” is selected as a correction candidate, and when the operator presses the key “2”, “candidate # 2” is selected as a correction candidate. The Similarly, when the operator depresses the key “0”, it is selected to re-input from the beginning of the character string in the input field of the erroneous input item 21c-1. Further, when the operator depresses the key “+”, the next erroneous input item is displayed.

  When either “candidate # 1” or “candidate # 2” is selected by the operator, the slip input apparatus 100 selects a character in the input field of the erroneous input item 21c-1 corresponding to the selected correction candidate. Modify the correction candidates. For example, in FIG. 5, when “candidate # 1” is selected by inputting the key “1”, the character string “1235001520” in the input field of the erroneous input item 21c-1 is the sixth and seventh digits from the top. Are replaced with “1235010520”.

  The correction to the correction screen 21c is also reflected in the input field of the input item of the input screen 21a corresponding to the input field of the erroneous input item 21c-1. Thereby, the operator can correct the character string in the input field of the erroneous input item 21c-1 without re-inputting all the character strings in the input field of the erroneous input item 21c-1.

  In addition, since the correction location is limited to the range of correction candidates, it is less likely to cause an error in correction input than when all the character strings are re-input. In addition, the number of correction operations is reduced and the time required for correction is reduced as compared with the case where only the portion where the input is changed is selected and corrected from the state where all the character strings can be re-input.

  If any of the displayed correction candidates is determined to be inappropriate, the operator can re-input from the beginning of the character string in the input field of the erroneous input item 21c-1 by selecting “re-input”. When the operator determines that the character string in the input field of the erroneous input item 21c-1 is not an erroneous input, the operator corrects the character string in the input field of the erroneous input item 21c-1 by selecting “Next”. Do not. When the character string in the input column is not an erroneous input, for example, there is a case where there is an error in OCR recognition.

  FIG. 5 shows an example in which the correction screen 21c is displayed as a separate window from the input screen 21a. However, the present invention is not limited thereto. For example, the operation item 21c-2 on the correction screen 21c may be displayed below the input field where the cursor is displayed on the input screen 21a.

  In addition, the slip input apparatus 100 can select each item of the operation item 21c-2 by key input. As a result, for example, the operator can input and correct input items by key input in the same manner as when entering letters of a slip, so that the work efficiency of correction increases.

FIG. 6 is a block diagram illustrating an example of functions of the slip input system. The recognition rate DB server 200 includes a recognition rate information storage unit 210 and a recognition rate determination unit 220.
The recognition rate information storage unit 210 stores a recognition rate table in which information related to the recognition rate is associated with each type of character capable of OCR recognition. The information related to the recognition rate includes a recognition rate for each character type and information for calculating the recognition rate.

  The recognition rate determination unit 220 receives information related to the recognition rate from each slip input device 100 for each type of OCR-recognizable character, and the recognition rate determination unit 220 receives information related to the received recognition rate. Aggregate for each type of OCR recognizable character. The recognition rate determination unit 220 calculates a recognition rate for each type of character that can be OCR-recognized, based on information related to the aggregated recognition rates. Then, the recognition rate determination unit 220 updates the recognition rate table with the information related to the aggregated recognition rates and the calculated recognition rate. The recognition rate determination unit 220 distributes the contents of the updated recognition rate table to each slip input device 100.

  The slip input device 100 includes a recognized character data storage unit 110, an input character data storage unit 120, a recognition rate information input / output unit 130, a recognized character data acquisition unit 140, an input character data acquisition unit 150, and a correction candidate determination unit 160.

  The recognized character data storage unit 110 stores a recognition rate table in which a recognition rate is associated with each type of character capable of OCR recognition. The recognized character data storage unit 110 stores recognized character data in which a recognition rate is associated with each character of the recognized character string.

The input character data storage unit 120 stores input character data input by an operator from each input field of the input screen 21a.
The recognition rate information input / output unit 130 receives information indicating the recognition rate for each type of character capable of OCR recognition from the recognition rate DB server 200 periodically (for example, at the start of every business), and stores the recognition rate table. Update. In addition, the recognition rate information input / output unit 130 aggregates information for calculating the recognition rate for each character that can be recognized by OCR, and the aggregated information is regularly recognized (for example, at the end of each business). Send to server 200.

  The recognized character data acquisition unit 140 receives document data from the scanner 20. The document data includes a character code of a character string (recognized character string) recognized by OCR. Information indicating any input item on the input screen 21a is associated with the character code of the character string included in the document data. The recognized character data acquisition unit 140 generates the input screen 21a based on the received document data. The recognized character data acquisition unit 140 displays the generated input screen 21 a on the display 21.

  The recognized character data acquisition unit 140 generates recognized character data by associating the recognition rate stored in the updated recognition rate table with each character included in the recognized character string. The recognized character data acquisition unit 140 stores the generated recognized character data in the recognized character data storage unit 110.

  The input character data acquisition unit 150 acquires the character code of the character string input by the operator in each input field on the input screen 21a as input character data. The input character data acquisition unit 150 stores the acquired input character data in the input character data storage unit 120. Further, the input character data acquisition unit 150 updates the display of the input field on the input screen 21a based on the correction candidate selected by the operator. Then, the input character data acquisition unit 150 updates the input character data corresponding to the updated input field.

  The correction candidate determination unit 160 compares the corresponding input character data and the recognized character data for each input item on the input screen 21a, and specifies an input item that may be erroneously input by the operator. It should be noted that the correction candidate determination unit 160 may display the input screen 21a with emphasis on the input item identified as having the possibility of being erroneously input. As a method of highlighting, for example, the color of the input field is changed.

The correction candidate determination unit 160 determines the presence / absence of a correction candidate based on the corresponding input character data and recognized character data for an input item that may be erroneously input.
The correction candidate determination unit 160 displays on the correction screen 21c an input item that is specified as having a possibility of being erroneously input. Moreover, the correction candidate determination part 160 displays a correction candidate on the correction screen 21c, when there exists a correction candidate. Furthermore, the correction candidate determination unit 160 accepts an operator's correction through the correction screen 21c for an input item that may have been erroneously input.

Next, tables used by the slip input system 5 will be described with reference to FIGS.
FIG. 7 is a diagram illustrating an example of a recognition rate table. The recognition rate table 111 is a table that stores information in which a recognition rate is associated with each character capable of OCR recognition. The recognition rate table 111 is stored in the recognized character data storage unit 110. The recognition rate table 111 is updated at the start of every business based on the recognition rate received from the recognition rate DB server 200.

  The recognition rate table 111 has items of characters and recognition rates. In the character item, information indicating a character type capable of OCR recognition is set. Information indicating the recognition rate of the character type is set in the item of recognition rate.

  FIG. 8 is a diagram illustrating an example of authentication character data and input character data. The recognized character data 112 is data in which a recognition rate is associated with each character included in the recognized character string. The recognized character data 112 is stored in the recognized character data storage unit 110.

  The recognized character data 112 has items of an input item, a recognition rate, and a character. In the input item field, information for identifying an input item input by the operator is set on the input screen 21a. In the item of recognition rate, information indicating the recognition rate corresponding to the characters recognized by OCR is set. The recognition rate is used when the correction candidate determination unit 160 determines correction candidates. In the character item, a character that has been OCR recognized by the scanner 20 is set. Hereinafter, a character string in which characters from the left end to the right end are arranged in order is referred to as a character string. For example, the character string of the recognized character data 112 is “123”.

  The input character data 121 is a character string input by the operator from each input field of the input screen 21a. The input character data 121 is stored in the input character data storage unit 120 for each input field of the input screen 21a. The input character data 121 has a character item. In the character item, a character input by the operator is set. Hereinafter, a character string arranged in order from the character located at the right end to the character located at the left end is referred to as a character string. For example, the character string of the input character data 121 is “133”.

  The correction candidate determination unit 160 compares the character strings of the recognized character data 112 and the input character data 121 that match the input item items. When one or more characters that do not match exist, the correction candidate determination unit 160 determines that there is a possibility that the input item corresponding to the input character data 121 has been erroneously input.

  For example, in FIG. 8, the recognized character data 112 and the input character data 121 are the second character from the left of the input character data 121 while the second character from the left of the recognized character data 112 is “2”. Is different because it is “3”. Therefore, the correction candidate determination unit 160 determines that the input item “amount” corresponding to the input character data 121 may be input incorrectly.

  FIG. 9 is a diagram illustrating an example of a recognition rate calculation table. The recognition rate calculation table 211 is a table that stores information in which information related to the recognition rate is associated with each type of character capable of OCR recognition. The recognition rate calculation table 211 is stored in the recognition rate information storage unit 210 of the recognition rate DB server 200. The recognition rate calculation table 211 is updated by the recognition rate DB server 200 at the end of every business.

The recognition rate calculation table 211 has items of characters, the number of times of recognition, the number of correct answers, and a recognition rate.
In the character item, a character type capable of OCR recognition is set.
The number of times of OCR recognition by each scanner 20 for the corresponding type of character is set in the item of the number of recognition times.

  In the correct answer item, the number of times of OCR recognition that is correct among the number of recognition times is set. Specifically, the recognition rate DB server 200 calculates the recognition rate based on the ratio of the correct number of OCR recognitions to the number of times of OCR recognition. For example, when the operator confirms the input, whether or not the answer is correct is compared with the character string of the recognized character data that matches the input item and the character string of the confirmed input character data. It is determined that there is.

  Next, the slip input device 100 will be described with reference to FIGS. Used in FIGS. In the recognized character data 112a to 112c and the input character data 121a to 121c, description of input items is omitted.

FIG. 10 is a diagram illustrating an example in which one correction candidate is determined.
It is statistically clear that there are many mistakes that the input order is changed as an erroneous input of an operator who is accustomed to the key input operation like an operator at a store window. This is considered to be an erroneous input that occurs because the speed of key input is too high.

  If the input order is incorrect, the input is incorrect for two adjacent character pairs in the input character string. Therefore, the correction candidate determination unit 160 searches for such a character pair from the input character string.

  In FIG. 10, the character string of the recognized character data 112a is “123146”, and the character string of the input character data 121a is “123164”. That is, it is assumed that the character string “1235146” is recognized by OCR recognition and the character string “1235164” is input by the operator. The recognition rate corresponding to each character of the character string of the recognized character data 112a is “95”, “90”, “85”, “85”, “95”, “90”, “60” from the top. .

  In this case, first, the correction candidate determination unit 160 compares the character strings of the recognized character data 112a and the input character data 121a. In this case, in the recognized character data 112a, “4” in the second digit from the lower order and “6” in the lowest order do not match the input character data 121a. Therefore, the correction candidate determination unit 160 determines that there is a possibility that an input item corresponding to the input character data 121a is erroneously input.

  Next, the correction candidate determination unit 160 selects, for the recognized character data 112a, a character whose corresponding recognition rate is equal to or greater than a threshold value from characters determined to be inconsistent. In FIG. 10, the recognition rate corresponding to “4” in the second digit from the lower order is “90”, and the recognition rate corresponding to the lowest order “6” is “60”. Therefore, when the threshold value is “90”, the correction candidate determination unit 160 selects “4” whose corresponding recognition rate is “90” or more. Regarding the selected “4”, since the certainty of recognition is more than a certain level, it is considered that there is a high possibility that the second digit from the lower order is erroneously input.

  Next, the correction candidate determination unit 160 selects, for the recognized character data 112a, a character whose recognition rate corresponding to a character adjacent to the selected character is less than a threshold value. In FIG. 10, the characters adjacent to “4” in the second digit from the lower order are “1” in the third digit from the lower order and “6” in the lowest order. The recognition rate corresponding to “1” is “95”, and the recognition rate corresponding to “6” is “60”. For this reason, the correction candidate determination unit 160 selects only “6” whose corresponding recognition rate is less than “90”.

  For the selected position “6”, since the certainty of recognition has not reached a certain level, whether it is erroneously input by the operator, erroneously recognized in OCR recognition, or correctly input. It is difficult to judge. However, there is a high possibility that an adjacent position, that is, the second digit from the lower order, is erroneously input. As described above, in view of the high probability of erroneous input that the order of input is switched, the second digit from the lower order where the probability of erroneous input is high, is adjacent to this, and accurate determination is impossible. It can be considered that there is a high possibility that a certain lowest order has been switched and input.

  On the other hand, for the third digit from the bottom, the certainty of recognition is above a certain level, and the character comparison results match, so it is highly probable that they were correctly input.

  For the above reasons, the correction candidate determination unit 160 determines the characters “6” and “4” of the input character data 121a corresponding to the pair of the second and lowermost characters specified from the character string of the recognized character data 112a. "Is identified as a pair of characters whose input order is switched. The correction candidate determination unit 160 determines that “46” in which the positions of the identified character pairs are exchanged is a correction candidate. “46” determined as a correction candidate is displayed as “candidate # 1” on the correction screen 21c.

  As a result, the operator can correct the erroneous input by selecting the correction candidate by pressing one key without re-inputting the character string “1235146” corresponding to the recognized character data 112a from the top. Correction can be made more efficient.

  In addition, in the process of determining a pair of characters whose input order has been switched by the above procedure, it is possible to newly determine that an erroneous input has occurred at a character position where the certainty of recognition has not reached a certain level. Can also be obtained.

  In the slip input device 100, the same threshold value (90%) is used for a threshold value for determining a character with a high recognition rate and a threshold value for determining a character with a low recognition rate. The value of may be used.

  FIG. 11 is a diagram illustrating an example in which two correction candidates are determined. In FIG. 11, the character string of the recognized character data 112b is “12335046”, and the character string of the input character data 121b is “12335064”. The recognition rate corresponding to the character string of the recognized character data 112b is “95”, “90”, “85”, “85”, “80”, “90”, “60” from the top.

  In this case, first, the correction candidate determination unit 160 compares each character of the recognized character data 112b and the input character data 121b, and the recognition rate corresponding to the recognized character data 112b from the characters determined to be inconsistent is equal to or greater than a threshold value. Select a character. In this case, as in the description of FIG. 10, when the threshold value is “90”, the correction candidate determination unit 160 selects “4” whose corresponding recognition rate is “90” or more.

  Next, the correction candidate determination unit 160 selects a character whose recognition rate corresponding to the character adjacent to the selected character is less than the threshold for the recognized character data 112b. In FIG. 11, the characters adjacent to “4” are “0” and “6”. The recognition rate corresponding to “0” is “80”, and the recognition rate corresponding to “6” is “60”. Therefore, the correction candidate determination unit 160 selects “0” and “6” whose corresponding recognition rates are less than “90”. That is, in this case, it is determined that there is a possibility that the input order is changed for the second and third digits from the lower order, and for the first and second digits from the lower order.

  Next, the correction candidate determination unit 160 identifies the character pair “4” with the selected recognition rate being equal to or greater than the threshold and the character pair “0” with the selected recognition rate being less than the threshold. Similarly, the correction candidate determination unit 160 specifies a character pair of a character “4” whose selected recognition rate is greater than or equal to a threshold and a character “6” whose selected recognition rate is less than the threshold. Then, the correction candidate determination unit 160 converts the character “0” and “6” pair and the “6” and “4” pair of the input character data 121b corresponding to the specified pair into the characters that are erroneously input. Identify the pair.

  Then, the correction candidate determination unit 160 determines “60” and “46” in which the positions of the identified pairs are exchanged as correction candidates. “60” determined as a correction candidate is displayed as a candidate # 1 on the correction screen 21c, and “46” determined as a correction candidate is displayed as a candidate # 2 on the correction screen 21c.

  Thus, when the recognition rates of characters adjacent to a character with a high recognition rate are both less than the threshold, there are two correction candidates. In this case, the operator can select either candidate # 1 or candidate # 2. In the description of FIG. 11, the operator can correct an erroneous input by selecting candidate # 2. Further, even when there are two correction candidates as described above, the operator can correct an erroneous input by selecting a correction candidate by pressing one key.

  FIG. 12 is a diagram illustrating an example in which correction candidates are not determined. In FIG. 12, the character string of the recognized character data 112 c is “1235146”, and the character string of the input character data 121 c is “1035164”. Also, the recognition rate corresponding to the character string of the recognized character data 112c is “95”, “90”, “85”, “85”, “95”, “90”, “60” from the left.

  In this case, first, the correction candidate determination unit 160 compares each character of the recognized character data 112c and the input character data 121c, and the recognition rate corresponding to the recognized character data 112c from the characters determined to be inconsistent is equal to or greater than a threshold value. Select a character. In this case, when the threshold value is “90”, the correction candidate determination unit 160 selects “0” and “4” whose corresponding recognition rate is “90” or more.

  Here, according to the above procedure, a plurality of characters whose recognition rate is equal to or higher than the threshold and which are determined to be inconsistent by the comparison result are detected, and a character pair that is considered that the input order is switched for each detected character is found. In this case, a method of correcting each detected character using the correction screen 21c can be considered. However, since the operator's key input is fast as described above, it is more likely that it will be faster to re-enter the entire character string that may have been erroneously input than from the beginning rather than performing correction by this method.

  Accordingly, the correction candidate determination unit 160 does not present a correction candidate when there are two or more characters having a corresponding recognition rate equal to or higher than a threshold in characters determined to have been erroneously input. In this case, the operator re-inputs the character string of the recognized character data 112c from the top. As a result, the time required for correcting the input item determined to be erroneous input can be shortened, and the correction work can be made efficient.

Next, with reference to FIGS. 13 to 15, processing when characters described in the slip are input by the operator will be described with reference to flowcharts.
FIG. 13 is a flowchart illustrating an example of processing for supporting character input. In the following, the process illustrated in FIG. 13 will be described in order of step number.

  (Step S11) When the operator sets a slip on the scanner 20, the scanner 20 recognizes the slip by OCR and generates document data. The scanner 20 transmits the generated document data to the slip input device 100.

  (Step S <b> 12) The recognized character data acquisition unit 140 receives document data from the scanner 20. The recognized character data acquisition unit 140 generates the input screen 21a based on the received document data. For example, the recognized character data acquisition unit 140 selects the character code corresponding to each input item on the template of the input screen 21a from the document data, and displays the corresponding character to generate the input screen 21a. The model of the input screen 21a is stored in advance in a storage area such as the recognized character data storage unit 110, for example. Then, the recognized character data acquisition unit 140 displays the generated input screen 21 a on the display 21.

  (Step S13) The recognized character data acquisition unit 140 generates recognized character data 112 by associating the recognition rate stored in the recognition rate table 111 with each character included in the recognized character string included in the received document data. To do. Specifically, the recognized character data acquisition unit 140 searches the recognition rate table 111 for the recognition rate corresponding to the character type for each character included in the recognized character string included in the received document data. Match the recognition rate to that character.

  (Step S14) The operator inputs the characters described in the slip into the corresponding input field on the input screen 21a. The input character data acquisition unit 150 acquires, as input character data 121, a character string input by the operator for each input field on the input screen 21a. Then, the process proceeds to step S21.

FIG. 14 is a flowchart (continued) illustrating an example of processing for supporting character input. In the following, the process illustrated in FIG. 14 will be described in order of step number.
(Step S21) The correction candidate determination unit 160 selects one input item.

  (Step S22) The correction candidate determination unit 160 acquires the input character data 121 corresponding to the selected input item from the input character data storage unit 120. The correction candidate determination unit 160 acquires the recognized character data 112 corresponding to the selected input item from the recognized character data storage unit 110. The correction candidate determination unit 160 compares the acquired character strings of the input character data 121 and the recognized character data 112.

  (Step S23) The correction candidate determination unit 160 determines whether the character string comparison results of the input character data 121 and the recognized character data 112 do not match. If the comparison results do not match, the process proceeds to step S24. If the comparison results match, the process proceeds to step S29.

(Step S24) The correction candidate determination unit 160 highlights and displays the selected input item on the input screen 21a.
(Step S25) The correction candidate determination unit 160 acquires a character whose recognition rate is equal to or greater than a threshold value from each character of the recognized character data 112 that does not match the corresponding character of the input character data 121.

  (Step S26) The correction candidate determination unit 160 determines whether or not the number of characters acquired in step S25 is one. If the acquired number of characters is 1, the process proceeds to step S27. If the acquired number of characters is other than 1, the process proceeds to step S28.

(Step S27) The correction candidate determination unit 160 determines whether or not there is a correction candidate for the character acquired in step S25 and having a recognition rate equal to or higher than a threshold value. Details will be described with reference to FIG.
(Step S28) The correction candidate determination unit 160 displays the correction screen 21c. At this time, “candidate # 1”, “candidate # 2”, and “re-input” are not displayed in the operation item 21c-2 of the correction screen 21c. Moreover, in the displayed correction screen 21c, the cursor is displayed at the head position of the input field of the erroneous input item 21c-1. From this state, the operator re-inputs the character string from the top into the input field of the erroneous input item 21c-1 by performing a key operation. Then, the operator selects the “next” item (presses the key “+”) from the operation item 21c-2 on the correction screen 21c, and displays the next erroneous input item.

  (Step S29) The correction candidate determination unit 160 determines whether all input items on the input screen 21a have been selected. If all input items have been selected, the process ends. If there is an unselected input item, the process proceeds to step S21.

  14 is performed after all the input items on the input screen 21a are input. As another example, every time an operator inputs an input field (for example, the cursor is moved to another input field). May move), the processing of steps S22 to S28 may be executed.

FIG. 15 is a flowchart illustrating an example of processing for determining a correction candidate. In the following, the process illustrated in FIG. 15 will be described in order of step number.
(Step S31) The correction candidate determination unit 160 acquires characters whose recognition rate is less than the threshold (90%) from the characters adjacent to the characters acquired in step S25 of FIG.

  (Step S32) The correction candidate determination unit 160 determines whether the number of characters acquired in step S31 is one. If the number of characters acquired in step S31 is 1, the process proceeds to step S33. If the number of characters acquired in step S31 is other than 1, the process proceeds to step S34.

(Step S33) The correction candidate determination unit 160 acquires one correction candidate according to the procedure shown in FIG.
(Step S34) The correction candidate determination unit 160 determines whether the number of characters acquired in step S31 is two. If the number of characters acquired in step S31 is 2, the process proceeds to step S35. If the number of characters acquired in step S31 is other than 2 (that is, 0), the process proceeds to step S38.

(Step S35) The correction candidate determination unit 160 acquires two correction candidates according to the procedure shown in FIG.
(Step S36) The correction candidate determination unit 160 displays the correction candidates on the correction screen 21c. At this time, when executing step S33, the correction candidate determination unit 160 displays “candidate # 1” but does not display “candidate # 2” in the operation item 21c-2 of the correction screen 21c. Moreover, the correction candidate determination part 160 displays "candidate # 1" and "candidate # 2" in the operation item 21c-2 of the correction screen 21c, when step S35 is performed.

  Then, the correction candidate determination unit 160 waits for the operator to make a selection input for the operation item 21c-2. If “candidate # 1” or “candidate # 2” is selected (key “1” or key “2” is pressed), the process proceeds to step S37. If “re-input” is selected (key “0” is pressed), the process proceeds to step S38.

  (Step S37) The correction candidate determination unit 160 corrects the input field of the erroneous input candidate based on the correction candidate selected by the operator. Then, the correction candidate determination unit 160 reflects the correction to the input field of the erroneous input item 21c-1 in the input field of the corresponding input item on the input screen 21a. Then, the operator selects the “next” item (presses the key “+”) from the operation item 21c-2 on the correction screen 21c, and displays the next erroneous input item. Thereby, the correction reflected in the input field is confirmed.

  (Step S38) The operator re-enters the character string from the beginning in the input field of the erroneous input item 21c-1. Then, the operator selects the “next” item (presses the key “+”) from the operation item 21c-2 on the correction screen 21c, and displays the next erroneous input item. Thereby, the correction reflected in the input field is confirmed.

  According to the slip input device 100 of the second embodiment, the correction candidate determination unit 160 compares the character strings of the recognized character data 112 and the input character data 121. Next, in the recognized character data 112, a character having a recognition rate equal to or greater than a threshold is acquired from a character string that does not match, and a character having a recognition rate less than the threshold is acquired from a character adjacent to the acquired character. Then, the correction candidate determination unit 160 acquires a character pair of the input character data 121 corresponding to a pair of a character that is greater than or equal to the acquired threshold and a character that is adjacent to the character and has a recognition rate that is less than the threshold.

  Since there is a high possibility that the character string pairs acquired in this way have been replaced and input by the operator, a pair in which the positions of the characters included in the acquired pairs are replaced can be determined as a correction candidate for erroneous input. Then, the correction candidate determination unit 160 presents the character string determined as the correction candidate on the correction screen 21c.

As a result, the operator can determine whether or not there is an erroneous input for an input item that may be erroneously input in consideration of the correction candidate, so that the determination as to whether or not there is an erroneous input can be made more quickly.
Further, the correction candidate determination unit 160, when selected by the operator from the presented correction candidates, corrects an input item that is erroneously input to the content of the selected correction candidate. Here, if only the erroneously input character is selected and corrected from the entire character string including the erroneous input, there is a possibility that an erroneous input such as a wrong digit will be made again. Therefore, in general, when there is an erroneous input in the character string of the input character data 121, the operator re-enters the character string from the top in the input field corresponding to the input character data 121. That is, even if only one character is erroneously input in the character string of the input character data 121, it is re-input for seven characters. On the other hand, the above processing makes it possible for the operator to make corrections by selecting from the proposed correction candidates, so that fewer characters need to be input than re-inputting all of them, making it easier to correct incorrect input. Is done. In addition, since the probability of erroneous input at the time of re-input can be suppressed, the accuracy of correction of erroneous input is increased.

  In the slip input system 5, instead of presenting correction candidates, the input screen 21a or the correction screen is set so that only characters determined to be erroneous input candidates among the character strings displayed in the input field can be corrected. 21c may be controlled. Thereby, the operator can suppress erroneous input such as a digit difference for characters other than the characters that are candidates for erroneous input. Therefore, the accuracy of correcting erroneous input is improved.

  The process of presenting a pair of correction candidates as described above is basically applied to a character string in which one character is input by operating one key. For example, it is desirable that the process of presenting correction candidates be applied to input items in which numbers are input. For example, the present invention can be applied to an input item in which alphabets are input. On the other hand, for example, for an input item to which a character string including Kanji characters is input, when the process of steps S21 to S23 is executed in the process of FIG. 14, the determination is “YES” in step S23. The process of step S28 should just be performed unconditionally.

  In the slip input system 5, the recognition rate is used as an example of the accuracy representing the accuracy of OCR recognition. However, for example, a parameter used during the OCR recognition process can be used as the accuracy representing the accuracy of the OCR recognition. For example, when character recognition is performed by collating an image read by the scanner 20 with a character pattern, the degree of coincidence with the pattern for the recognized character may be used as the accuracy of OCR recognition.

  When the parameter used during the OCR recognition process is used as the OCR recognition accuracy, the slip input apparatus 100 acquires the parameter from the scanner 20. In addition, when the slip input device 100, not the scanner 20, performs OCR recognition, the slip input device 100 itself can output the parameters.

Further, for example, a value obtained by correcting the parameter used during the OCR recognition process by the above-described recognition rate can be used as the accuracy of the OCR recognition.
As described above, the processing of the character input support device 10 according to the first embodiment can be realized by causing the character input support device 10 to execute a program. The processing of the slip input device 100 and the recognition rate DB server 200 according to the second embodiment can be realized by causing each device to execute a program. Such a program can be recorded on a computer-readable recording medium (for example, the recording medium 23). As the recording medium, for example, a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like can be used. Magnetic disks include FD and HDD. Optical discs include CD, CD-R (Recordable) / RW (Rewritable), DVD, and DVD-R / RW.

  When distributing the program, for example, a portable recording medium in which the program is recorded is provided. It is also possible to store the program in a storage device of another computer and distribute the program via the network 30. The computer stores, for example, a program recorded on a portable recording medium or a program received from another computer in a storage device (for example, HDD 103), and reads and executes the program from the storage device. However, a program read from a portable recording medium may be directly executed, or a program received from another computer via the network 30 may be directly executed. Further, at least a part of the information processing described above can be realized by an electronic circuit such as a DSP (Digital Signal Processing), an ASIC, or a PLD (Programmable Logic Device).

1 image 2, 3 character string 10 character input support device 11 acquisition unit 12 determination unit

Claims (7)

On the computer,
The first character string recognized from the image, the accuracy indicating the recognition accuracy of each character included in the first character string, and the second character string input in response to the input operation are acquired. And
Each character included in the first character string and each character included in the second character string are compared on a one-to-one basis from the beginning, the compared characters are different, and the corresponding accuracy is the first Extracting a first character that is greater than or equal to a threshold from the first character string;
Extracting, from the first character string, a second character that is adjacent to the first character and the corresponding accuracy is less than a second threshold;
Determining a character pair of the second character string corresponding to the first character and the second character as an erroneous input;
A character input support program characterized by causing processing to be executed. In the computer,
Displaying an input screen for accepting a correction input for the second character string, and limiting a correctable region of the second character string to the region of the character pair in the input screen;
The character input support program according to claim 1, further comprising executing processing. In the process of displaying the input screen,
Replacing the order of the characters included in the character pair, displaying the character pair after the replacement as a correction candidate on the input screen,
Confirming the input of the second character string in which the order of the characters included in the character pair is changed in accordance with the input operation of the correction confirmation;
The character input support program according to claim 2, wherein: In the extraction of the second character, when a character that is adjacent to the first character and the corresponding accuracy is less than a second threshold is extracted from both sides of the first character, the input screen is displayed. In the process to display
Of the characters included in the second character string, one extracted character and the first character pair corresponding to the first character, and the other extracted character and the first character The second character pair is displayed on the input screen as a correction candidate in a state where the order of the characters is changed,
In response to an input operation for selecting one of the displayed correction candidates, input of the second character string in which the order of characters included in the first character pair or the second character pair is changed is performed. Determine,
The character input support program according to claim 2, wherein:   In the extraction of the first character, the input screen is displayed when the compared characters are different from each other and a plurality of characters whose corresponding accuracy is the first threshold value or more are extracted from the first character string. 5. The character input support program according to claim 2, wherein in the processing, the entire second character string is set as an area that can be corrected on the input screen. 6. In the computer,
The order of the characters included in the second character string is changed, and the changed character pair is displayed on the screen as a correction candidate.
The character input support program according to claim 1, further comprising executing processing. The first character string recognized from the image, the accuracy indicating the recognition accuracy of each character included in the first character string, and the second character string input in response to the input operation are acquired. An acquisition unit to
Each character included in the first character string and each character included in the second character string are compared on a one-to-one basis from the beginning, the compared characters are different, and the corresponding accuracy is the first A first character that is equal to or greater than a threshold is extracted from the first character string, and a second character that is adjacent to the first character and has a corresponding accuracy that is less than a second threshold is the first character. A determination unit that extracts from the character string and determines that the character pair of the second character string corresponding to the first character and the second character is an erroneous input;
A character input support device comprising:

Images