JP2015118488A - System, method and program for inputting account data - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simplify recognition processing of date, a total amount and the like and reduce false recognition in account data input.SOLUTION: An accounting process support system includes: format determination means for determining a format of an original voucher from a scanned image; character recognition means; and a first character recognition dictionary containing kanji characters and the like; and a second character recognition dictionary for numerical recognition. The accounting process support system determines a receipt by format determination, uses the first character recognition dictionary to find a character string containing a kanji character and the like, and uses the second character recognition dictionary to recognize an assumable numeral portion. For recognition of date and a total amount, date is specified using a date keyword, and a numerical portion of the specified date character string is recognized placing priority on numerical recognition. A total amount is specified using a total amount keyword, and a total amount character string is specified by a relative position of the obtained total amount character string, and is recognized placing priority on numerical recognition. If a total amount cannot be specified using a keyword, a total amount character string is specified based on a character size and thickness, and is recognized placing priority on numerical recognition.

Description

  In the accounting processing system used when accounting software is used in a consulting company of an accounting office or when journal entry is input at an accounting office, such as a receipt, a passbook, etc. The present invention relates to a technology for supporting data input from primaries.

  Conventionally, certified public accountant offices and tax accountant offices (hereinafter simply referred to as “accounting offices” or “offices”) receive data and primordial certificates from the consultant in various formats. Accounting for the advisor. In recent years, with the spread of personal computers, basic materials submitted from consultants to accounting firms are often electronic media.

  When submitting basic materials from an advisor to an accounting firm in electronic media (data), generally enter the transaction data using accounting software (such as a cash register form) at the advisor, and keep the data in the network. To the accounting office or stored in a storage medium such as a memory card and passed to the accounting office.

  Data input to such accounting software is usually input from the keyboard while the user visually observes the original certificate such as a receipt or a passbook generated during the transaction. When inputting data, there is a technique for reducing input work from a keyboard and reducing data input mistakes (for example, Patent Document 1).

  In addition, for example, in the case of a consultant who is not familiar with personal computer operations and difficult to input data into accounting software, he / she enters handwritten characters on a standard form preprinted in the past, and the completed form is sent to the accounting office. There is a method of submitting and OCR processing a slip at an accounting office to improve data input efficiency (for example, Patent Document 2).

  In addition, the original vouchers such as receipts and passbooks generated at the time of transactions were acquired as an image by an optical device such as a scanner, and numbers and the like were extracted from the image to determine the certainty of the determination result with a dictionary or the like. There is a technique for improving the accuracy of character recognition as compared with a threshold (for example, Patent Document 3).

  In addition, when the amount of money such as receipts and passbooks generated during transactions is read with an optical device such as a scanner and the result is converted to text data, the character at which position is the amount Since it is unknown, there is a technique for specifying the meaning of the character string of the read text by specifying the type of the read form (for example, Patent Document 4).

  In addition, the original certificate such as receipts and passbooks generated at the time of transaction is acquired as an image by an optical device such as a scanner, and numbers etc. are cut out from the image, and the word candidates closest to the characters such as the cut out numbers are dictionaryd. (For example, Patent Document 5).

JP 10-275196 A JP-A-8-30719 JP-A-11-224305 JP 9-330363 A JP-A-5-40854

  When using a pre-printed standard OCR read-only slip, the user enters the letters on the slip by hand while looking at the original certificate such as receipts and passbooks generated at the time of the transaction at the business partner. Will do. This is a cumbersome operation, and there is a problem that many erroneous inputs occur.

  When reading original receipts such as receipts or passbooks with a scanner, OCR processing, and outputting data to accounting software, if the original identification documents to be read are standardized and supported by accounting software Although the recognition accuracy increases and the data input work can be reduced, the format of the receipt (receipt) is not standardized because the issuer can set it arbitrarily, and the bankbook also has a different lineup of items depending on the financial institution It has not been. For this reason, there is a problem that a receipt or passbook in a format not supported by the accounting software becomes unreadable or the recognition accuracy is greatly reduced.

  In addition, in order to cause the original certificate to be read by the scanner and to perform the OCR process, the process of identifying which part of the original certificate and outputting the data becomes complicated. In fact, the receipt has a problem that it is difficult to grasp where the data to be taken in as accounting information is located because the format is not defined and the meaning is not given to the description, and therefore, it is difficult to extract necessary information. The complexity of the process is a factor that increases the number of places where misrecognition occurs, and the user has more opportunities to correct the misrecognized places. Therefore, there is a problem that the data input work cannot be reduced.

  By the way, since the date and the total amount are important as information related to the accounting processing obtained from the original certificate, when the original certificate is read by the scanner and the OCR process is performed, the information of the date and the total amount is obtained with high accuracy. Is important.

  Therefore, the present invention is a process in a case where an image obtained by reading an original certificate such as a receipt or a passbook with a scanner is subjected to OCR processing and output to accounting software at an accounting firm's consulting company or the like. In addition to simplifying the recognition process of date and total amount, etc., the purpose is to reduce input errors from the keyboard and reduce data input mistakes when inputting data by reducing misrecognition.

In order to achieve the above object, the present invention recognizes a format determination means for determining the format of a primitive voucher from a read image obtained by reading a primitive voucher, and recognizes and recognizes characters from the read image using a character recognition dictionary. In an accounting processing support system comprising character recognition means for generating candidate text data, a first character recognition dictionary including kanji characters and a second character recognition dictionary for number recognition are used as the character recognition dictionary. Prepared,
If the image is a receipt,
It is assumed that the character recognition means uses the first character recognition dictionary to recognize the number of characters using the second character recognition dictionary in addition to the recognition candidate text data as a character string including kanji and the like. Generate text data of recognition candidates,
Date information extraction means for extracting date data character strings as date information from the text data of the recognition candidates;
A total amount information extracting means for extracting a total amount data character string as total amount information from the text data of the recognition candidate,
The date information extracting means extracts a character string including a date keyword related to specification of date information from the recognition candidate text data as significant information for accounting processing, and estimates the character string as date information data. , Performing text data optimization of recognition candidates that preferentially handle recognition candidates when the estimated character string is assumed to be a number related to date information,
If there is a character string that includes a total amount keyword related to the identification of the total amount information from the text data, the total amount information extraction unit uses the character string in a specific relative positional relationship with the character string for accounting processing. If the character string including the total amount keyword is not present in the text data, it is determined based on the character size or the thickness of the character. Is extracted, and the character string including the extracted number is estimated as the total amount information,
The recognition candidate text data optimization that preferentially treats the recognition candidate when the estimated character string is assumed to be a number related to the total amount information is performed.

  In one embodiment, the total amount information extraction unit sets the right side as the specific relative positional relationship when there is a character string including a total amount keyword related to the specification of the total amount information from the text data, The number on the right side of the total amount keyword of the character string is determined as the total amount.

  In one embodiment, the first character recognition dictionary is constituted by a recognition dictionary having minimum definition data narrowed down to characters used in a primitive voucher for accounting processing.

  In one embodiment, when the second character recognition dictionary is assumed to be a number when the character string to be recognized is assumed to be a number, when performing character recognition processing assuming that the number is a character string, It is composed of a recognition dictionary that is removed to improve the recognition rate of numbers.

  In one embodiment, the date keyword includes at least one of “year”, “month”, “day”, “/”, “.”, “−”.

  In one embodiment, the total amount keyword is one of “total”, “current total”, and “purchase total”.

  In the present invention, at a client company or an accounting office, a receipt or a passbook is read with a scanner or a mobile device with a camera function, and OCR processing is performed, and data is output to accounting software. Therefore, it is possible to reduce input work from the keyboard and to reduce data input mistakes. The present invention can find necessary accounting data such as a date and a total amount, even for a free format primitive voucher, and flow it into an input screen of accounting software.

  Also, as dictionary data in character recognition processing, using dictionary data with only numbers as recognition candidates, or using dictionary data narrowed down to characters used in primitive vouchers, the processing speed and recognition accuracy can be improved. Can be achieved.

FIG. 1 is a schematic diagram illustrating an overall outline of an accounting data input support system according to the present invention. FIG. 2 is a diagram for explaining date information extraction processing of the system when the original voucher is a receipt. FIG. 3 is a diagram for explaining the total amount information extraction process of the system when the original voucher is a receipt. FIG. 4 is a diagram for explaining the processing of the system when the original voucher is a bankbook. FIG. 5 is a diagram showing a system configuration of the accounting data input support system and its variations. FIG. 6 is a flowchart for explaining the overall processing of the method according to the present invention. FIG. 7 is a flowchart for explaining step S06-04 in FIG. 6 in more detail. FIG. 8 is a flowchart for explaining the post-processing (S06-05 to 06) of FIG. 6 in the case of performing character recognition of date information on receipts and receipts. FIG. 9 is a flowchart for explaining the post-processing (S06-05 to 06) of FIG. 6 when character recognition is performed on the total amount information of receipts and receipts. FIG. 10 is a flowchart for explaining the details of the character recognition process for transaction information in the bankbook. FIG. 11 is a diagram illustrating a structure example of a character recognition dictionary. FIG. 12 is a diagram illustrating an example of a keyword dictionary. FIG. 13 is a diagram illustrating an example of the estimation processing table.

  EMBODIMENT OF THE INVENTION The form for implementing this invention is demonstrated in detail below, referring an accompanying drawing. FIG. 1 is a schematic diagram illustrating an overall outline of an accounting data input support system according to the present invention. The present invention relates to a free format receipt (receipt), a passbook, etc., as a scanner, a digital camera capable of photographing a subject and storing it as image data, a digital video, etc. (function for uploading images by wireless function) Mobile phone with camera, smartphone with camera, mobile terminal with camera, tablet terminal with camera, mobile terminal such as mobile camera (portable terminal device such as mobile phone and smartphone, camera) Accounting software that operates on the computer 100 by reading the image of the camera of a tablet terminal with function (hereinafter simply referred to as “mobile device with camera function”) or other reading device 20 that can digitize the original voucher, and performing image processing The accounting data is automatically reflected in It is intended to be. Since the what is described in advance which parts if slip form accounting software is compatible is known, can be reflected in the accounting data performs OCR process reads the necessary part However, unformatted free format primitive vouchers cannot do this. In view of this, the present invention seeks to search for necessary accounting data such as date and total amount, even in a free format primordial voucher, as shown in the enlarged screen on the lower side, and to flow it into the input screen of accounting software. Features.

  First, the system configuration of the accounting data input support system according to the present invention will be described with reference to FIG. In one embodiment, the accounting data input support system 1 is configured as a computer system installed in a consultant or accounting firm. As shown in FIG. 5A, the system 1 includes a reading device 20 such as a scanner for reading the original voucher 10 or a mobile device with a camera function, and a terminal device connected to the reading device 20 directly or via a network. 100 and so on. The terminal device 100 includes an input unit 110 such as a keyboard and a mouse, a display unit 120 such as a display, an output unit 130 such as a USB port and a portable storage drive, and a communication unit 140 for connecting to a network such as the Internet. A control unit 150 that controls various controls of the terminal device 100 and a storage unit 160 that stores various data are provided.

  The storage unit 160 includes at least a primitive voucher image data storage unit 161, a character recognition dictionary 162, a word dictionary 163, a keyword dictionary 164, a transaction information storage unit 165, and an accounting data storage unit 166. The functions of the control unit 150 include a reading processing unit, an image processing unit, a management unit, a character recognition processing unit, a dictionary handling unit, a recognition candidate generation unit, a word matching unit, an optimization processing unit, a transaction data generation unit, and a journal processing. There is a department. These functions are realized by reading out and developing accounting software or a separate program module stored in the storage unit 160 by the control unit 150. The embodiment of the present invention is not limited to the form shown in FIG. 5-A, and some elements may be omitted or other elements may be included.

  The reading processor reads the original voucher with the reading device 20 and creates image data (read image). This reading processing unit is indicated by a broken line because the reading of the original certificate is not necessarily performed by the terminal device, but other devices having a camera function (mobile devices with a camera function, digital cameras, etc.) It is because it may be performed in. That is, the reading processing unit (illustrated with a broken line) is included in the terminal device 100 or the reading device 20, and another terminal device 100a connected to the network, the terminal device 20, a data server, NAS (Network Attached Storage), etc. May be. In addition to the desktop computer, the terminal device 100 may be a notebook computer or a mobile device with a camera function. In the case of a mobile device with a camera function, the reading device (scanner or the like) shown in FIGS. 1 to 5A is configured inside the terminal device 100.

  The image processing unit includes a format judging means, and from the image data (read image) created by the reading processing unit, whether the voucher type of the image data is a receipt (receipt) or the like, or is a bankbook A rough judgment is made, or the format judgment means for judging a rough format such as where the character string of the original voucher such as a receipt or a passbook is located.

  The management unit associates the image data created by the reading processing unit with the journal data input by the journal processing unit.

  The character recognition means is realized by cooperation of a character recognition processing unit, a dictionary handling unit, a recognition candidate generation unit, and an optimization processing unit. Specifically, using the character recognition dictionary selected by the dictionary handling unit (a dictionary including kanji, a number dictionary), the character recognition processing unit calculates a feature amount of each character and compares it with the character recognition dictionary. The recognition candidate generation unit generates a recognition candidate in consideration of the comparison result, detects keywords related to dates and totals using the keyword dictionary selected by the dictionary handling unit, and the optimization processing unit performs predetermined processing based on the keywords and the like. The estimation candidate is adjusted by increasing the priority as a recognition candidate and optimizing the recognition candidate by the optimization processing program based on the estimation.

  The date information extraction means is realized by cooperation of a dictionary handling unit, a recognition candidate generation unit, and an optimization processing unit. Specifically, the optimization processing unit estimates a character string including a date keyword related to specifying date information as date information data from the recognition candidate text data output by the recognition candidate generation unit of the character recognition unit, and the estimated character string , The recognition candidate text data optimization is performed to increase the priority of the recognition candidate.

  The total amount information extraction unit operates by the cooperation of the dictionary handling unit, the recognition candidate generation unit, the word matching unit, and the optimization processing unit. Specifically, the word collation unit collates the recognition candidate text data output from the recognition candidate generation unit of the character recognition unit with words such as the sum of the word dictionary selected by the dictionary handling unit, and relates to the identification of the total amount information. When there is a character string including the total amount keyword, the optimization processing unit estimates a character string having a specific relative positional relationship with the character string as total amount information, and the character string including the total amount keyword is If it does not exist in the text data, the optimization processing unit estimates the character string including the number with the largest character size as the total amount information, and the recognition candidate when the estimated character string is assumed to be the number related to the total amount information. Optimize text data for recognition candidates to increase priority.

  The journal processing unit creates journal data from the input contents of a two-screen accounting input screen as shown in FIG. 1, for example, and stores it in the accounting data storage unit.

  The transaction data generation unit saves a summary, which is transaction information, from the journal data generated by the journal processing unit in the transaction information storage unit.

  Further, the image data storage unit 161, the transaction information storage unit 165, and the accounting data storage unit 166 of the storage unit 160 are in a data server or NAS (Network Attached Storage) in the terminal device 100 or connected to the network. Also good. Modification examples of such a system configuration are shown in FIGS. In the embodiment of FIG. 5B, the image data storage unit 161 is arranged on the server device 50 connected to the network, and the terminal device 100b is different from the terminal device 100a connected to the reading device 20 that has read the original voucher. Accounting data input support processing is performed. Further, in this example, the image data storage unit 161 is arranged in another terminal device 100a without providing the server device 50, and this is read out by the terminal device 100b connected to the network to perform the accounting data input support processing. Is also possible.

  In addition, since the original voucher usually relates to a transaction that has occurred at an advisor, the terminal device 100b and the other terminal device 100a are both assumed to be terminal devices on the advisory side. The device 100a may be a terminal on the accounting office side or a terminal device brought in by a staff of the accounting office. This is because, as a service on the accounting office side, the original voucher may be deposited from the consultant and read. Or, conversely, the terminal device 100b may be a terminal device on the accounting office side, and the other terminal device 100a may be configured to read the original voucher as a terminal device on the advisory side. Moreover, both the terminal device 100b and the other terminal device 100a may be configured as a terminal device on the accounting office side.

  In the embodiment of FIG. 5C, image data is stored in the server device 50 directly from a mobile device with a camera function such as a smartphone (that is, the reading device 20) via the network, and is read and processed by the terminal device 100. Is called. The server device 50 in this case may be a NAS or a cloud. As described above, the system configuration is diverse and not intended to be particularly limited. It is only necessary that an image obtained by reading the primitive voucher exists in some form. The present invention will be described below with a focus on how the character recognition processing related to the primitive voucher has been devised starting from the fact that there is an image obtained by reading the primitive voucher.

  Next, an accounting data input support method when the original voucher is a receipt according to the first embodiment will be described with reference to FIGS. Whether the type of the original voucher is a receipt or a bankbook includes keywords such as “receipt” and “receipt” by the above-described format determination means (corresponding to the layout analysis in step S06-02 in FIG. 6). (In the case of a receipt) or whether the keyword such as “Balance” is included (in the case of a bankbook), whether the character strings are miscellaneous (in the case of a receipt), or in the vertical direction However, the details are omitted because it is sufficient to use a general technique. The receipt of a general restaurant is shown on the left side of FIG. This receipt is not fixed for each store, but usually stores information such as store information such as store name and telephone number, date, individual order contents and amount, total amount, deposit amount, and change. In the present invention, attention is paid to the date and the total amount (displayed as “current total” in the figure) which are important as accounting data. FIG. 2 is a diagram showing an overview of processing when attention is paid to date information, and FIG. 3 is a diagram showing an overview of processing when attention is paid to total amount information.

  First, with reference to FIG. 2, an outline of processing when attention is paid to date information will be described. The image data read by the reading device 20 is subjected to OCR processing to generate a recognition candidate character string, and then a keyword (for example, year, month, day, /,-,. Period (English month notation such as Nov)) (Similar processing is hereinafter referred to as keyword detection processing) (S101), and when there is a keyword related to date, a series of character strings including the keyword is converted to date This character string is extracted by presuming that it is information (S102). Next, the process proceeds to optimization processing of the character string.

  In the example of FIG. 2A, the actual receipt describes “September 6, 2013”, and the system estimates that this portion is date information from the read image in step S102. However, depending on factors such as receipt status and reading accuracy setting, this part may be recognized as “2DJ8 / 09 / Q” or “× + −year O or 6 units” by primary reading. The optimization process is performed in consideration of such a case. Even in the latter case of the previous sentence, the character string “year” is included, so that the character string can be extracted as the date / time information.

  In the optimization process when it is assumed that it is date information, since the date information is composed of numbers, the recognition candidates when it is assumed that all the characters included in the character string are numbers are used (S102). ), Recognition candidate optimization processing is performed (S103). That is, the character recognition dictionary is fixed to, for example, only the numbers shown on the right side of FIG. 11 so that characters other than numbers are not candidates. In addition, a value calculated by extracting a plurality of feature points for each character from the image data and calculating the feature amount is compared with the feature amount of the dictionary data to determine a number having the highest similarity. If none of the numbers match the feature points to some extent, a recognition error occurs. As a result, an accurate number is extracted from the estimated character string, and the recognized text data becomes “2013? 9? 6?" ((B) of S103).

  Here, the “?” Part is a part corresponding to a kanji such as year, month, day, etc., and represents a case where it cannot be determined as a number recognition candidate (assuming that there is no candidate with high similarity). Corresponding characters are separately supplemented with character string recognition candidates such as date and time that have been recognized in the normal recognition dictionary (step S103 (b)), and the date "September 6, 2013" is completed. You may make it make it. In this way, a predetermined estimation process is performed according to the detection of the keyword, and as a process on the assumption that a character string related to a number such as a date is included, a recognition dictionary for numbers is assumed assuming that the number is a number. Recognition accuracy of date information can be improved by generating a recognition candidate by performing character recognition used and preferentially handling a recognition candidate that seems to be a number when it is assumed to be a number. In addition, the keyword detection of step S101 can use what performed word collation about the recognition candidate produced | generated by the character recognition using a normal recognition dictionary. In addition, the recognition process using the number dictionary when it is assumed to be a number may be performed in advance before the date information keyword is detected, or may be performed after the date information keyword is detected. Also good.

  Here, the processing for preferentially handling the recognition candidates that are likely to be numbers is not limited to the above example, and the following processing may be performed. FIG. 2-B is a reference diagram for explaining a modification of the recognition candidate optimization process. In particular, when a character recognition process is performed on a receipt using a normal recognition dictionary without using recognition candidates generated by recognizing characters using a recognition dictionary for numbers on the assumption that the number is a character, the system uses multiple characters for each character. Characters with matching feature points are listed as candidates by comparing feature values generated by extracting feature points with dictionary data. In this process, as shown on the left side of FIG.

  Next, in the character recognition of the date part, the priority of the “candidate with the largest number” is increased without using the recognition candidate (a in S103) when it is assumed to be a number as shown in FIG. May be optimized. More specifically, in the example of FIG. 2-B, for example, “2DJ, September Q day 8”, “XX + year O or 6 units”, “2013” If the candidate "9 or 6 days a year" is listed, the conventional system recognizes "2DJ8 September Q day", which has the highest match, as the final candidate. In the present invention, this part is estimated to be date information. Therefore, the priority of the candidate “9th or 6th 2013” with the largest number is raised to be the most promising candidate.

  With reference to FIG. 3, the optimization process regarding reading of the total amount of receipts and receipts will be described. The system searches for keywords related to the total amount from the OCR data of the receipt shown on the left side of FIG. 3-A (S201). In one embodiment, the keyword related to the total amount includes character strings such as “total”, “current total”, “purchase total”, and is registered in advance in a keyword dictionary (see FIG. 12) in the system. . When such a keyword is detected, it is estimated that the character string on the right side of the keyword is a number related to the total amount (S202). At this time, since the character string of the total amount often includes characters such as “¥” and “yen”, it is determined whether or not the character string on the right side of the keyword includes these characters. The total amount may be estimated.

  In the keyword detection process for checking whether there is a character string that matches the keyword registered in the keyword dictionary, a recognition candidate generation process for each character is performed as a recognition candidate in advance, and a plurality of characters are generated using the word dictionary. By performing collation at the combined word level, the accuracy as a recognition candidate is improved, and a recognition candidate is searched (see step S06-05 in FIG. 6).

  It is sufficient that all the numbers of the total amount are clearly read in the processing up to S202, but recognition candidates specialized for numbers may be further optimized (S203). That is, character recognition is performed by applying the number-only dictionary shown on the right side of FIG. 11 to the estimated character string so as to increase the number recognition accuracy. More specifically, a recognition candidate when a number is assumed is generated in advance, or when a keyword indicating the total amount is detected, a recognition candidate when a character string on the right side is assumed as a number is shown. The recognition accuracy is increased by generating and using the recognition candidates with priority. In the example of FIG. 3A, in step S203 b), the recognition candidate when it is assumed to be a number is “? 4? 200”, and the character part of the number is preferentially used as a candidate and recognized. The “?” Part, which is a part with low accuracy, is replaced with the part “¥”, “,” other than the number, which is a part that is a recognition candidate when it is a kanji character, etc. Get “¥ 4200”. Further, it is possible to improve the recognition accuracy that the character string is money amount information by determining whether there is a number after the “¥” character or before the “yen” character.

  Next, as an example shown in FIG. 3B, a detection process as an alternative means in place of keyword detection when a keyword related to the total amount does not exist in the receipt will be described. As in this example, when there is no character string related to the total amount in the original voucher, the size of each character is recognized in image reading and OCR processing. In normal OCR processing, the characters in the image are simply converted into text and the size of the characters is not focused. For example, in the receipt, the total amount may be represented as the largest number. By comparing the sizes, it is estimated that the largest number is the total amount (S204). In this case, since the store name and the like are often displayed in large size, the recognition accuracy can be improved by using only numbers as comparison targets. In addition, since the phone number and date may be displayed in large size, in order to exclude this, for example, a string of characters such as “4 digits-4 digits” is estimated as the total amount assuming that there is a high possibility of a phone number. The character string including the date keyword (year, month, day, /, etc.) may not be estimated as the total amount even if the font size is large. Thereafter, optimization processing similar to that described above is performed on the estimated numeric character string, and a recognition result is output. Also, in this embodiment, the description has focused on extracting the largest character as important information. However, the thickest character string, the italicized character string, or the character string with a different font type is used as important information. You may make it extract.

  Information on the date and the total amount detected in this way is passed to the accounting software side and is automatically reflected on the two-screen accounting input screen shown in FIG. As shown in the screen example at the bottom of Fig. 1, a primordial voucher (in this case, a receipt) is displayed on the left side of the screen, and the date information and total amount included in this are automatically reflected on the input screen on the right side. . The user confirms that the date and the total amount are the same by referring to the original certificate, and if necessary, automatically inputs the date information and the amount information such as the total amount, etc. It can be corrected by touch input. In particular, for date information, various date information such as the date of issue and date of transaction (purchase date, etc.) or point expiration date may be printed on the receipt. It is also possible to display and select a plurality of recognition candidates for date information. Alternatively, depending on the company's accounting process, the settlement date may be entered instead of the transaction date of the receipt. In this case, in addition to manual correction, the data entry date is automatically reflected. You may do it.

  In addition, accounting information is created by inputting other information (store name, who you were with, purpose, payment method, etc.). In this example, the input screen is a question format based on 5W1H (when, where, with whom, how, and what), and the user can create accounting data by directly entering the question answer selection it can. The details of the question-type slip are not characteristic features of the present invention, but are disclosed in detail in other applications of the applicant of the present application, and therefore description thereof is omitted here. The accounting data created in this way is stored in the accounting data storage unit 166 of the system. When the accounting data is created on the side of the advisor, it is stored in a storage medium from the output unit 130 and transferred to the accounting office, or sent from the communication unit 140 to the accounting office via the network.

  In addition, although the above demonstrated using the example that a primitive voucher is a receipt of a restaurant, this invention is applied also to primitive vouchers, such as a receipt, an invoice, and the payment of various taxes other than a receipt. can do. The processing method in that case is the same as described above.

  The details of the data input support method from the primitive voucher will be described with reference to FIGS. FIG. 6 is a flowchart for explaining the overall processing of the method according to the present invention. First, an image of a slip (primary voucher) read by the reading device 20 is captured (S06-01). As preprocessing, a character string is cut out from the image (S06-02), and then cut out in character units (S06-03). In the cutout of the character string, it is determined where the character string (that is, a cluster of dots) is in the image, and in the cutout of one character unit, pitch determination such as full-width / half-width is performed. As a cutting method used in this case, a projection method is known in which fine dot clusters constituting a character string are sampled from the X-axis side and the Y-axis side to form a histogram to grasp the location where the dot clusters exist. It has been. Note that in this pre-processing, it may be possible to determine a document type of a large frame such as a receipt or a passbook (see step S07-01 in FIG. 7).

  Next, character recognition is performed for each character string included in the image as recognition processing (S06-04). This processing is realized by character recognition processing, dictionary handling, and recognition candidate generation. In this case, as described above, in addition to character recognition in the normal mode (recognition candidate generation using a recognition dictionary including kanji and the like), in order to perform character recognition assuming that the character string is a number as necessary, If a number-only dictionary is applied to generate recognition candidates that are assumed to be numbers, the recognition accuracy can be improved. Note that, as described above, the recognition candidate generation process when it is assumed to be a number may be executed each time when it is estimated that the information is date information or (total) amount information, or in advance, all characters You may make it perform about a column.

  Next, word recognition is performed at the word level for each character recognized in units of characters, and recognition candidate adjustment is performed as a process for improving accuracy as a recognition candidate of a character string corresponding to a keyword such as “total”. Is performed (S06-05). If a character string such as “Total” is not found even though it is known that the voucher type is “Receipt”, etc., by the keyword search for the month and day, or the total amount, the largest numeric string (1) circle number 1 and (2) circle number 1 in step S06-06 (for example, for date information, step S101 in FIG. 2-A, total) The money amount information corresponds to step S201 in FIG. 3-A or keyword detection / substitution processing in FIG. 3-B). Through this process, a notable character string such as a date or a total amount is determined.

  Then, optimization processing is performed on the character string to be noted (circle numbers 2-3 in (1) and circle numbers 2-3 in (2) in S06-06). That is, for a character string that is estimated as a date or total amount by keyword detection, if there is a keyword, an optimization process assuming that it is a number is performed, and if there is no keyword, focus on the character string with the largest number, for example. The optimization process is performed assuming that this is a number. In this way, the optimum recognition candidate is determined and output (S06-07).

  FIG. 7 is a diagram for explaining step S06-04 in FIG. 6 in more detail. First, the document type of the read image is determined (S07-01). This is because, for example, the format of an OCR slip that is supported by accounting software is registered in the system in advance, and it is possible to determine which slip format by reading the identifier in the image, or to check the passbook from the size of the image and the layout of the frame. It is determined by determining whether it is a slip (primary voucher) that cannot be specified in other formats. In this way, the document type may be automatically determined, or the user may input by looking at the voucher image by presenting a selection screen such as a passbook, a receipt or a receipt. If it is an OCR slip that is supported by the system (S07-02: Y), it is already known where it will be written, so the system automatically reads the necessary information and automatically displays the accounting input screen. (S07-12). This processing itself is a known technique, and detailed description thereof is omitted in this specification.

  When the read image is a slip (primary certificate) whose format cannot be specified, first, character recognition is performed using a dictionary including kanji and the like (S07-03 to 07). FIG. 11 shows an example of the structure of the character recognition dictionary. In FIG. 11, the left side shows the registered contents of the character recognition dictionary including kanji and the right side shows the registered contents of the character recognition dictionary for numbers. As shown in FIG. 11, a character code, a feature amount, and the like are registered for each character. The character recognition dictionary for numbers contains only numbers, or only numbers and mathematical symbols. Kanji and kana are not registered, so the amount of information is small. In this case, it is possible to effectively avoid misrecognition of numbers. On the other hand, a character recognition dictionary including kanji characters may be a general OCR dictionary, but the present invention is related to accounting data input support, and thus specializes only in characters that are likely to be used in primitive vouchers. It may be configured. As a result, even when a dictionary including kanji and the like is used, the processing speed can be increased and the recognition rate can be improved.

  Returning to FIG. 7, in step S07-03, a character recognition dictionary (namely, the dictionary of FIG. 11) for a slip whose format cannot be specified (that is, the dictionary of FIG. 11) is acquired, and first a character recognition dictionary including kanji is set. (S07-04), character recognition processing is performed (S07-05-07). Next, in order to perform character recognition when it is assumed to be a number, a character recognition dictionary for numbers is set (S07-08), and a character recognition dictionary is performed (S07-09 to 11). In this way, the character recognition of the original voucher image is performed. When generating recognition candidates that are assumed to be numbers, it is possible to generate recognition candidates with accuracy, so the character string including date information, the right side of the character string such as the sum or the largest number string is estimated as the total amount. However, it is possible to improve recognition accuracy by preferentially handling a recognition candidate when it is assumed that the number is assumed and then preferentially handling it. Note that, as described above, the recognition candidate generation process when it is assumed to be a number may be executed each time when it is estimated that the information is date information or (total) amount information, or in advance, all characters You may make it perform about a column.

  FIG. 8 is a flowchart for explaining the post-processing (S06-05 to 06) of FIG. 6 in the case of performing character recognition of date information on receipts and receipts. As described above, keywords for extracting important information (for example, date and total amount) in accounting data are registered in the system in advance. An example of this keyword dictionary is shown in FIG. As shown in this figure, keywords such as “total”, “current total”, “purchase total”, “receipt amount”, “payment”, etc. are registered as keywords indicating total amounts, Keywords such as “day, month, year”, “/”, “−”, “.”... Are registered. As shown in FIG. 12, the priorities such as first, second,... Are indicated for the keywords. This is related to, for example, date information, and the receipt has points in addition to the issue date and transaction date. It is difficult from the viewpoint of character recognition to determine which information is significant as accounting information, including multiple date information such as expiry date of the date (it is not necessarily clearly stated as issue date or transaction date) ), Statistically analyzing receipts and handling character strings extracted in the above priority order as transaction dates. Can be used as Similarly, priorities are set for keywords such as totals, and are used for efficiently discriminating character strings that are likely to be numbers of total amounts by appropriately weighting them. The system sets a keyword dictionary related to the date (S08-01), and performs a keyword search for the original voucher for which character recognition has been performed (S08-02). When the keyword is hit here, that is, for example, when a character such as “month, day” is detected in a certain character string (S08-03: Y), an estimation process according to the keyword (step S102 in FIG. 2) is performed (step S102 in FIG. 2). S08-04).

  In the estimation process in step S08-04, in one embodiment, for example, the process of the estimation process table shown in FIG. 13 is performed according to the extracted keyword. That is, for each type of voucher, an estimation process is dynamically selected for each recognition target, and an optimization process linked to the selected estimation process is dynamically selected. More specifically, as shown in FIG. 13, for example, when a keyword related to the date information of the receipt is extracted, the optimization processing program ID: P002 is selected and “estimates the part other than the date as a number” Then, the optimization process as described with reference to FIG. 2 is performed. In this case, the accuracy of estimation may be improved by taking into consideration whether or not the character immediately before “year”, “month”, and “day” includes a number recognition candidate. Then, the optimization process described with reference to FIG. 2 is performed according to the keyword (S08-05). That is, as shown in FIG. 2-A, the recognition candidate for the part that is assumed to be a number is “numbers” in consideration of both the recognition candidates for numbers and the recognition candidates for kanji characters. Character recognition in a normal dictionary without using recognition candidates in the case of numbers, as shown in FIG. 2-B. Among the recognition candidates that have been subjected to the above, the priority order of candidates that include a large number may be increased. Furthermore, excluding other recognition candidates, assuming that there are 4 digits (or last 2 digits) before "Year" and 2 digits (or 1 last digit) before "Month, Day". Alternatively, it may be determined that the character string is not date information. In addition, since the primordial voucher for which journal entries are entered at accounting firms or consultants is usually in the near past, it is necessary to check whether the recognized date falls within a certain range (for example, five years ago or later). Judgment may be made to determine the certainty of determination and the certainty of extraction processing. In addition, when the keyword regarding the date information of step S08-03 is not found, the user himself / herself may be input on the two-screen accounting input screen shown in FIG.

  FIG. 9 is a flowchart for explaining the post-processing (S06-05 to 06) of FIG. 6 when character recognition is performed on the total amount information of receipts and receipts. Similar to FIG. 8, keyword types are set, and keyword detection is performed with reference to the keyword dictionary (FIG. 12) (S09-01 to 03). For example, if there are keywords related to the total amount such as “current total” and “total”, the estimation process and optimization process are executed according to the keyword, such as the character string (number string) on the right side of the keyword is used as the total amount. (S09-04 to 05), recognition candidates in character string units are generated (S09-06). On the other hand, when a keyword related to the total amount information is not found by keyword search (S09-03: N), important information replacing the keyword is detected as an alternative process (S09-08). That is, as shown in FIG. 13, when there is no keyword as the total amount information of the receipt, the largest number string is estimated as the number of the total amount (S09-09, optimization processing program ID: P011). More specifically, the size of characters included in the original certificate image is compared in units of one character, or the size is compared in units of cut out recognized character strings, leaving a large character string group, and finally remaining The largest character string is found by extracting a group of character strings or by calculating an average value of the entire character string and extracting and comparing several large character clusters. At this time, since the total amount is a number, it is possible to exclude a store name or the like by detecting only the character string of the number. Furthermore, when only one large character string is found, it is estimated that the character string is total amount information, or when three large character strings are found, for example, the character on the right side from the center A string or a character string at the uppermost position may be estimated as total amount information. This is because it is assumed that there are a total amount, a deposit amount, and a change as numbers that are displayed relatively large in the receipt, and the total amount is often described at the top. When the total amount information is estimated in this way, recognition candidates are optimized (S09-10), and recognition candidates in character string units are generated (S09-06). In the above description, the example in which the size of the character string (character size) is grasped as the important information has been described. However, the thickness of the character (alone or superimposed on the size of the character) is considered. Also good.

  In addition, character size comparisons may be based on both the height and width (area) of the cut-out character, considering only the vertical height of the character or only the width in the horizontal direction. Alternatively, the determination may be made by weighting any element of the height, width, area, and the like. Regarding the determination of the height, width, and area of the cut-out character, it is usual to use an image after correcting the inclination at the time of reading, but the image before correction can be used. Needless to say, in order to compare the feature values with the recognition dictionary, the character size may be adjusted and normalized, but the comparison of the character size etc. is in the state before normalization. Do.

  Next, an accounting data input support method when the original voucher is a bankbook according to the second embodiment of the present invention will be described with reference to FIGS. Even when the original voucher is a bankbook, the processing of the flowchart of FIG. 6 is basically performed. The processing flow of FIG. 10 is a detailed processing of step S06-06 of FIG. It is. For example, as shown in Fig. 4-A, the bankbook is divided into fields such as "date", "payment", "deposit", and "balance" for each transaction, but the preprinted column names are printed lightly. However, there are cases where the “payment” and “deposit” fields are reversed by the financial institution, which makes it difficult to handle as a standard primordial voucher. Therefore, the present invention assumes that the left end is “date” and the right end is “balance”, and “payment” and “deposit” between them are automatically determined and can be taken in as accounting data. It is characterized by.

  In the passbook reading image shown in FIG. 4-A, as a premise, for example, in the OCR recognition process, whether or not there is a character string that matches a character string of a keyword such as date, balance, deposit, etc. If a character string that matches these keywords is found by performing a search, the number character string below will be understood as the contents of the corresponding keyword (date, balance, payment, deposit, etc.) It is reflected in the corresponding part of the formula accounting input screen (Fig. 1). On the other hand, even though it is known that the voucher type is “passbook”, the character string such as “balance” disappears due to noise removal processing or density adjustment as preprocessing, If no matching character string is found, the character string is cut out in line units (S301), the leftmost character string is estimated to be “date”, the rightmost character string is assumed to be “balance”, and the two at the center are “payment”. And “deposit” (S302: refer to the estimation processing table in FIG. 13). And the priority of the recognition candidate as a number is raised with respect to these character strings, and a candidate is adjusted (S303). The recognition candidate optimization process for adjustment is the same as the process in step S103 in FIG. 2-A for date information and the process in step S203 in FIG. Thereby, as shown in S304 of FIG. 4-A, the recognition result in which the date and the balance are fixed is output, but at this stage, it cannot be determined whether the two character strings in between are payment or deposit.

  Therefore, the present invention further determines the trend of increase / decrease in the balance of the preceding and succeeding lines (S403) as shown schematically in FIG. 4-B, and if the balance has increased from the previous line, the date of the line and the balance The number in the middle is a deposit, and conversely, if the balance is decreasing, it is recognized that the payment is made, and the content of the column is finalized (S404). In this case, it is determined whether or not the actual increment (30,000 yen in the example of step S403 in FIG. 4-B) matches the amount of money “30,000” in the left column (second column from the right). Then, recognition accuracy may be improved by adding verification. The keyword detection / substitution process S401 is the same as the process in step S301 in FIG. 4-A, and the estimation process S402 is the same as the process in step S302 in FIG.

  FIG. 10 is a flowchart for explaining the details of the character recognition process for transaction information in the bankbook. When a keyword serving as a headline is not found in character recognition of the image data of the passbook, important information replacing the keyword is detected as an alternative process (S10-01). For example, it is estimated that it is “date” with respect to the leftmost character string, and it is estimated that it is “amount” otherwise (S10-02). At this time, it may be estimated that the right end is “balance” and the two in the center are “payment” and “deposit”, respectively. Next, an optimization process according to important information is performed for each row to generate recognition candidates (S10-03 to 05).

  Next, after estimating that the rightmost character string is “balance” in each line, for example, pay attention to any non-empty line, determine the increasing / decreasing tendency of the balance in the upper and lower lines, and pay from the contents And the deposit line are determined (S10-06-09). In this way, the contents of the header column of the bankbook are confirmed, the meaning of the character string described below is confirmed, and the system automatically extracts the date and amount according to the contents of the transaction and accounts for it. It can be reflected on the input screen.

  In this way, the original voucher such as a receipt is different from the OCR slip where the description position and contents are determined by a frame line etc., and it is difficult to distinguish numbers / kanji and the recognition rate is low. Since it is information of date and total amount, these keywords are detected and estimated according to the keywords (if keyword is not found, keyword substitution is performed and important information is estimated) By adding optimization processing (adjustment processing such as preferential handling of recognition candidates and increasing the rank as a candidate), sufficient recognition accuracy specialized in recognition of dates and total amounts can be obtained.

  In addition, when character recognition that is assumed to be a number is performed at a predetermined stage, it is necessary to remove a kanji character and utilize a number character recognition dictionary specialized for numbers through a predetermined estimation process. In a series of processes that preferentially deal with recognition candidates when it is assumed to be numbers, assuming that it seems to be a number, the “date of transaction” and (total etc. It is useful in improving the recognition rate of “numbers such as monetary information)”. In addition, because the format of receipts etc. is not defined and meaning is not given to the description, it is difficult to grasp and extract what data you want to import as accounting information, but items (dates and journals) you want to import as accounting data By automatically extracting the amount of money (such as the total amount) to be entered, it is possible to facilitate the input.

  As described above, the embodiments and examples of the present invention have been described in detail. However, the technical scope of the present invention is not limited to the above-described embodiments and examples, and the present invention is limited to the appended claims. Various modifications and changes can be realized without departing from the scope of the invention, and it should be understood that all such modifications and changes belong to the technical scope of the present invention.

  INDUSTRIAL APPLICABILITY The present invention can be used for a computer system that uses accounting software in a consulting company of an accounting office or an accounting processing system that is used when journal entry is input at an accounting office.

DESCRIPTION OF SYMBOLS 1 Accounting data input support system 10 Primitive voucher 20 Reading apparatus 50 Server apparatus 100 Terminal apparatus 110 Input part 120 Display part 130 Output part 140 Communication part 150 Control part 160 Storage part 161 Image data storage part

Claims (12)

A format determining unit that determines a format of the original voucher from a read image obtained by reading the original voucher; and a character recognition unit that recognizes a character from the read image using a character recognition dictionary to generate recognition candidate text data. In the accounting processing support system comprising: a first character recognition dictionary including kanji and the like, and a second character recognition dictionary for number recognition, as the character recognition dictionary,
If the image is a receipt,
It is assumed that the character recognition means uses the first character recognition dictionary to recognize the number of characters using the second character recognition dictionary in addition to the recognition candidate text data as a character string including kanji and the like. Generate text data of recognition candidates,
Date information extraction means for extracting date data character strings as date information from the text data of the recognition candidates;
A total amount information extracting means for extracting a total amount data character string as total amount information from the text data of the recognition candidate,
The date information extracting means extracts a character string including a date keyword related to specification of date information from the recognition candidate text data as significant information for accounting processing, and estimates the character string as date information data. , Performing text data optimization of recognition candidates that preferentially handle recognition candidates when the estimated character string is assumed to be a number related to date information,
If there is a character string that includes a total amount keyword related to the identification of the total amount information from the text data, the total amount information extraction unit uses the character string in a specific relative positional relationship with the character string for accounting processing. If the character string including the total amount keyword is not present in the text data, it is determined based on the character size or the thickness of the character. Is extracted, and the character string including the extracted number is estimated as the total amount information,
An accounting data input support system, wherein text data optimization of a recognition candidate that preferentially handles a recognition candidate when the estimated character string is assumed to be a number related to total amount information is performed.   The total amount information extraction means sets the right side as the specific relative positional relationship when the character string including the total amount keyword related to the specification of the total amount information exists from the text data, and the character string 2. The accounting data input support system according to claim 1, wherein a number on the right side of the total amount keyword is determined as a total amount.   The said 1st character recognition dictionary is comprised by the recognition dictionary which has the minimum definition data narrowed down to the character used with the primitive voucher for accounting processing, The said 1 or 2 characterized by the above-mentioned. Accounting data input support system.   The second character recognition dictionary removes character strings other than numbers when performing character recognition processing assuming that the character string to be recognized is a number when the character string to be recognized is a number. The accounting data input support system according to any one of claims 1 to 3, wherein the accounting data input support system comprises a recognition dictionary for improving a recognition rate.   The date keyword includes at least one of "year", "month", "day", "/", ".", "-". The accounting data input support system as described in one.   2. The accounting data input support system according to claim 1, wherein the total amount keyword is any one of “total”, “current total”, and “purchase total”. A format determining unit that determines a format of the original voucher from a read image obtained by reading the original voucher; and a character recognition unit that recognizes a character from the read image using a character recognition dictionary to generate recognition candidate text data. In an accounting support system comprising
The character recognition dictionary includes a first character recognition dictionary including kanji and the like, and a second character recognition dictionary for number recognition,
If the image is a passbook,
It is assumed that the character recognition means uses the first character recognition dictionary to recognize the number of characters using the second character recognition dictionary in addition to the recognition candidate text data as a character string including kanji and the like. Generate text data of recognition candidates,
Date information extraction means for extracting date data character strings as date information from the text data of the recognition candidates;
A monetary amount information extraction means for extracting monetary amount data character string as monetary amount information from the text data of the recognition candidate,
The date information extraction means extracts character string data in line units from the text data as its constituent elements, further decomposes the character string data into item name units for each column, and generates a passbook data character string related to the date. With passbook date information extraction means,
The amount information extraction means extracts character string data in line units from the text data as its constituent elements, further decomposes the character string data into item name units for each column, and generates a passbook data character string relating to the amount With passbook amount information extraction means,
When the passbook date information extraction unit extracts character string data in units of rows, the character string in the leftmost column is extracted as positionally significant information for accounting processing, and the character string is date information. Estimating and performing text data optimization of recognition candidates that preferentially handle recognition candidates when the estimated character string is assumed to be a number related to date information,
Further, the passbook amount information extracting means extracts the rightmost character string as positionally significant information for accounting processing, estimates the character string as balance information, and uses the estimated character string as balance information. An accounting data input support system for generating the passbook data character string by optimizing text data of a recognition candidate that preferentially handles the recognition candidate when it is assumed to be such a number.   When the passbook amount information extraction unit extracts character string data in line units from the text data, it extracts an arbitrary line including a non-empty line above, the balance information of the extracted arbitrary line, 9. The accounting data input support system according to claim 7, wherein the payment information and the deposit information of the passbook data character string are discriminated by comparing the balance information of the lines.   The passbook amount information extracting means compares the balance information (A) included in the arbitrary row with the balance information (B) included in the upper row, and if (A)> (B), 9. The accounting data input support system according to claim 8, wherein an item next to balance information is determined to be deposit information, and the passbook data character string is generated.   10. The accounting data input support system according to claim 9, wherein the passbook amount information extraction unit generates the passbook data character string by regarding the remaining items of the passbook data character string as payment information. A format determining unit that determines a format of the original voucher from a read image obtained by reading the original voucher; and a character recognition unit that recognizes a character from the read image using a character recognition dictionary to generate recognition candidate text data. In the accounting data input support method in a computer system comprising the first character recognition dictionary including kanji and the like as the character recognition dictionary and the second character recognition dictionary for number recognition,
If the image is a receipt,
It is assumed that the character recognition means is a number using the second character recognition dictionary in addition to text data of recognition candidates as a character string including kanji and the like using the first character recognition dictionary. Generating text data of recognition candidates;
A date information extraction means, extracting the date data character string as date information from the text data of the recognition candidate as positionally significant information and extracting the character string;
A total amount information extraction means, extracting the total amount data character string as the total amount information from the recognition candidate text data as significant information for accounting processing, and extracting the character string;
When the date information extraction means estimates a character string including a date keyword related to the specification of date information from the text data of the recognition candidate as date information data, and assumes that the estimated character string is a number related to date information The step of optimizing the text data of the recognition candidate to handle the recognition candidate with priority,
When the total amount information extraction unit includes a character string including a total amount keyword related to the specification of the total amount information from the text data, the character string having a specific relative positional relationship with the character string is calculated as the total amount information. When the character string including the total amount keyword is not present in the text data, a unique character string is extracted based on the character size or character thickness, and the character string including the extracted number is totaled. Estimating the amount information, and performing the text data optimization of the recognition candidate preferentially handling the recognition candidate when the estimated character string is assumed to be a number related to the total amount information, Accounting data input support method. A format determining unit that determines a format of the original voucher from a read image obtained by reading the original voucher; and a character recognition unit that recognizes a character from the read image using a character recognition dictionary to generate recognition candidate text data. In the accounting data input support method in a computer system comprising the first character recognition dictionary including kanji and the like as the character recognition dictionary and the second character recognition dictionary for number recognition,
If the image is a receipt,
It is assumed that the character recognition means is a number using the second character recognition dictionary in addition to text data of recognition candidates as a character string including kanji and the like using the first character recognition dictionary. Generating text data of recognition candidates;
A date information extraction means, extracting the date data character string as date information from the text data of the recognition candidate as positionally significant information and extracting the character string;
A total amount information extraction means, extracting the total amount data character string as the total amount information from the recognition candidate text data as significant information for accounting processing, and extracting the character string;
When the date information extraction means estimates a character string including a date keyword related to the specification of date information from the text data of the recognition candidate as date information data, and assumes that the estimated character string is a number related to date information The step of optimizing the text data of the recognition candidate to handle the recognition candidate with priority,
When the total amount information extraction unit includes a character string including a total amount keyword related to the specification of the total amount information from the text data, the character string having a specific relative positional relationship with the character string is calculated as the total amount information. When the character string including the total amount keyword is not present in the text data, a unique character string is extracted based on the character size or character thickness, and the character string including the extracted number is totaled. Estimating the amount information, and performing the text data optimization of the recognition candidate preferentially handling the recognition candidate when the estimated character string is assumed to be a number related to the total amount information, Accounting data input support program.

Images