JP2022019445A - Image processing apparatus, method, and program - Google Patents

Abstract

To extract character strings of a character recognition result while dividing the character strings at an appropriate position.SOLUTION: An image processing apparatus stores in advance regular expression definition in which a regular expression formula describing a character string to be extracted and a parameter related to a space character defined in advance for every regular expression formula are associated with each other. The image processing apparatus inserts the space character into a character string obtained by executing character recognition processing on a document image based on a parameter related to the space character to create a character string for searching, searches for a portion matching the regular expression formula in the created character string for searching, and divides a character string resulting from the character recognition based on the portion searched for.SELECTED DRAWING: Figure 5

Description

The present invention relates to image processing devices, methods and programs.

There is a business of scanning paper documents and storing them electronically. Conventionally, there has been a system that recognizes characters and uses them for file names when digitizing. For example, there is a system in which a user selects a character recognition result from a document image and saves the character recognition result as a file name in an arbitrary storage. However, since the character recognition result is used, when the character recognition result fluctuates, for example, when an extra blank character exists in the character string to be set as the file name, the file name also contains the blank character, which is preferable. do not have. Therefore, Patent Document 1 discloses a method of converting a character recognition result into a character string suitable for a file name, such as removing a leading blank character when using the file name.

Japanese Unexamined Patent Publication No. 2013-74609

In the method of Patent Document 1, when a blank is included in the selected character recognition result, it can be converted into a character string suitable as a file name, but the range of the character string of the character recognition result selected as the file name is preferable in the first place. Sometimes not. For example, even if you want to select a date as a file name, one character string may be formed including the characters attached to the left and right of the date, and the entire character string may be selected. For example, if "payment date" is described as the item name of the date and the item name and the date cannot be separated, they are selected as one character string. In this case, the user needs to manually select and delete an extra part from the character brocade result of the selected character string.

The image processing apparatus of the present invention has a storage means for storing a regular expression definition in which a character string to be extracted is associated with a regular expression expression and a parameter related to a space character defined in advance for each regular expression expression.
Search characters by inserting space characters into the character string of the character recognition result by the character recognition means based on the character recognition means that executes character recognition processing for the document image and the parameters related to the space characters. Based on the generation means for generating a column, the search means for searching for a part matching the regular expression from the search character string generated by the generation means, and the part searched by the search means. It is characterized by having a dividing means for dividing a character string of a character recognition result.

According to the present invention, the character string of the character recognition result can be divided and extracted at an appropriate position.

It is a figure which shows the system configuration of an image processing system. It is a figure explaining the hardware composition of the image forming apparatus 101. It is a figure explaining the hardware configuration of the image processing server 102, and the user terminal 103. It is a figure which shows the example of the form image 400 and the character recognition result. It is a figure which shows the processing flow of 1st Embodiment. It is a figure which shows the list of the regular expression definition in 1st Embodiment. It is a table which shows an example of a character recognition result. It is an example of the result after the division process in the 1st embodiment. It is a figure which shows the list of the regular expression definition in the 2nd Embodiment. This is an example of the character recognition result in the second embodiment. It is a figure which shows the list of the regular expression definition in 3rd Embodiment. This is an example of the character recognition result in the third embodiment.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. It should be noted that the embodiments do not limit the present invention, and not all the configurations described in the embodiments are indispensable means for solving the problems of the present invention.

<First Embodiment>
FIG. 1 is a diagram showing a configuration example of the image processing system 100 according to the first embodiment. The image processing system 100 includes an image forming apparatus 101, an image processing server 102, and a user terminal 103. The image forming apparatus 101, the image processing server 102, and the user terminal 103 are connected to each other by the network 104 and can communicate with each other.

The image forming apparatus 101 receives and prints a print request (printing data) for image data from the user terminal 103, reads the image data with the scanner provided in the image forming apparatus 101, and the image data read by the scanner. It is a multifunction device that can print images. Further, the image processing server 102 is an image processing device capable of executing image processing described later on the image data read by the scanner of the image forming apparatus 101 and transmitting the image processing result to the user terminal 103. Is. The image processing server 102 may be a cloud, that is, a virtual server located on the Internet. The user terminal 103 can perform additional processing interactively with the user by an application provided with a user interface on the image processing result received from the image processing server 102. In the present embodiment, the user terminal 103 is assumed to be a general PC provided with a display, a keyboard, and a mouse, but may be, for example, a mobile terminal provided with a touch panel.

In the present embodiment, the image forming apparatus 101 scans a paper form such as an invoice, the image processing server 102 extracts necessary information from the form and stores it electronically, and the user terminal 103 confirms the extraction result. A series of data input support processes that provide a user interface that can be modified will be described.

FIG. 2 is a diagram showing an example of the configuration of the image forming apparatus 101. The image forming apparatus 101 includes a controller 201, a printer 202, a scanner 203, and an operation unit 204. The controller 201 includes a CPU 211, a RAM 212, an HDD 213, a network I / F 214, a printer I / F 215, a scanner I / F 216, an operation unit I / F 217, and an extended I / F 218.

The CPU 211 controls the entire image forming apparatus 101. The CPU 211 can control the exchange of data with the RAM 212, HDD 213, network I / F 214, printer I / F 215, scanner I / F 216, operation unit I / F 217, and extended I / F 218. Further, the CPU 211 expands the control program (instruction) read from the HDD 213 into the RAM 212, and executes the expanded instruction in the RAM 212. The HDD 213 stores a control program that can be executed by the CPU 211, set values used by the image forming apparatus 101, data related to processing requested by the user, and the like. The RAM 212 has an area for temporarily storing an instruction read from the HDD 213 by the CPU 211. Further, the RAM 212 can also store various data necessary for executing an instruction. For example, in image processing, the CPU 211 can perform processing by expanding the input data to the RAM 212.

The network I / F 214 is an interface for performing network communication with the device in the image processing system 100. The network I / F 214 can transmit the data reception to the CPU 211 and can transmit the data on the RAM 212 to the network 104. The printer I / F 215 can transmit the print data transmitted from the CPU 211 to the printer 202, and can transmit the state of the printer received from the printer 202 to the CPU 211. The scanner I / F 216 can transmit the image reading instruction transmitted from the CPU 211 to the scanner 203, transmit the image data received from the scanner 203 to the CPU 211, and transmit the state received from the scanner 203 to the CPU 211. Is. The operation unit I / F 217 can transmit an instruction from the user input from the operation unit 204 to the CPU 211, and can transmit screen information for the user to operate to the operation unit 204. The extended I / F 218 is an interface that enables an external device to be connected to the image forming apparatus 101. The extended I / F 218 includes, for example, a USB (Universal Serial Bus) type interface. The image forming apparatus 101 can read data stored in the external storage device and write data to the external storage device by connecting an external storage device such as a USB memory to the extended I / F 218. It is possible.

The printer 202 can print the image data received from the printer I / F 215 on paper and transmit the state of the printer 202 to the printer I / F 215.

The scanner 203 can read the information displayed on the paper placed on the reading unit, digitize it, and transmit it to the scanner I / F 216 according to the image reading instruction received from the scanner I / F 216. Further, the scanner 203 can transmit its own state to the scanner I / F 216.

The operation unit 204 is an interface for causing the user to perform an operation for giving various instructions to the image forming apparatus 101. For example, the operation unit 204 includes a liquid crystal screen having a touch panel, provides an operation screen to the user of the image forming apparatus 101, and accepts an operation from the user.

FIG. 3A is a diagram showing an example of the configuration of the image processing server 102. The image processing server 102 has a CPU 301, a RAM 302, an HDD 303, and a network I / F 304. The CPU 301 controls the entire image processing server 102. The CPU 301 can control the exchange of data with the RAM 302, the HDD 303, and the network I / F 304. Further, the CPU 301 expands the control program (instruction) read from the HDD 303 into the RAM 302, and executes the expanded instruction in the RAM 302.

FIG. 3B is a diagram showing an example of the configuration of the user terminal 103. The user terminal 103 has a CPU 311, a RAM 312, an HDD 313, a network I / F 314, and an input / output I / F 315. The CPU 311 controls the entire user terminal 103. The CPU 311 can control the exchange of data with the RAM 312, the HDD 313, the network I / F 314, and the input / output I / F 315. The display 320 is composed of a display device such as a liquid crystal display, and displays display information received from the input / output I / F 315. The input device 330 is composed of a pointing device such as a mouse or a touch panel, and a keyboard, receives an operation from a user, and transmits operation information to the input / output I / F 315. The HDD 313 can store the image processing result received from the image processing server 102 via the network I / F 314. In the present embodiment, the CPU 311 expands the application program read from the HDD 313 into the RAM 312, and the operation unit I / F 315 displays the display information and accepts the user operation.

FIG. 4A is a diagram showing an example of a form image 400 assumed in the present embodiment. The form image 400 is an image acquired by reading a paper document (for example, an invoice) with a scanner of the image forming apparatus 101. The item values 401 to 403 are examples of item character strings to be extracted by the image processing system 100. The item value 401 in FIG. 4A is the value of the title indicating the content of this document, the item value 402 is the value of the date indicating the issue date, and the item value 403 is the value of the billing amount. For the sake of explanation, the positions of the item values 401 to 403 are shown by a rectangular frame, but it is assumed that the rectangular frame is not described in the actual form image.

FIG. 4B is an example of a character string (OCR character string) of the character recognition result obtained when the general-purpose area analysis process and the optical character recognition (OCR) process are executed on the form image 400. .. Eight character areas of character strings 410 to 417 are specified, and an OCR character string is extracted from each character area. The position of each character area extracted based on the results of the area analysis process and the OCR process is shown by a rectangular frame. The character string 410 is obtained as a character string corresponding to one character area including the character string of the item value 401 and the character string on the left side thereof. Further, the character string 411 is extracted as a character string corresponding to one character area including the item value 402 and the character string on the left side thereof. Further, the character string 413 is also extracted as a character string corresponding to the area including the item value 403 and the character string on the left side thereof.

A use case for using this character string in the UI for creating a file name by the user will be described. For example, regarding a UI (user interface) in which when a user clicks a desired position on a form image, a character area corresponding to the clicked position is selected based on the area analysis result of FIG. 4 (b). explain. In such a UI, when the user clicks on the character string of "invoice", the character string of the character area based on the area analysis result (that is, the character string "ABC Co., Ltd. invoice" of the character area 410). Will be selected. Therefore, if the user wants to select only the document title part of the "invoice" as the file name, it is necessary to delete the extra "ABC Co., Ltd." character string from the selected character string. be.

Therefore, in the present embodiment, when the user selects a character string to be used as a file name, a character string such as item values 401 to 403 is appropriately extracted from the OCR character string as shown in FIG. 4 (b). The image processing to be performed will be described.

Extraction of the item value in the present embodiment indicates a state in which the target value is an independent character string to which no extra character is attached. For example, one character string "invoice" having an item value of 401 is independent of the form image 400.

The regular expression definition list 600 in FIG. 6 is an example showing a plurality of regular expression definitions in a table format. The regular expression definition in the present embodiment is defined by associating a combination of a regular expression expression and a regular expression parameter with each definition ID. The plurality of predefined regular expression definitions are stored in the HDD 303 of the image processing server 102. The regular expression expression describes the item to be extracted, for example, a character string to be extracted, such as a date, a telephone number, an amount of money, and a character included in a document title, in one regular expression expression. The regular expression parameter is a parameter defined for each regular expression expression and how to interpret the target OCR character string when performing a regular expression search. For example, a parameter describes how far the adjacent characters should be to be treated as a space character (blank character).

In the example of the regular expression definition list 600 of FIG. 6, three regular expression definitions 610, 620, and 630 are defined.

The regular expression definition ID 610 includes a regular expression pattern of "¥ S * book" and a regular expression parameter of "space = 2h". The regular expression pattern of "\ S * book" represents a pattern that combines multiple characters other than space characters (\ S) and the character "book", for example, "invoice", "quote", etc. The character string of can be searched as the corresponding pattern. The regular expression parameter "space = 2h" is a space if the distance between adjacent characters is at least twice the character height (h) when converting the OCR character string to the search character string. Indicates that characters are inserted and handled. In this embodiment, as a regular expression parameter, the character height is specified as a threshold value for treating as a space character. For example, the pixel size of an image, the physical distance on a paper surface, and the average character width are specified. Etc. may be used as a reference.

The regular expression definition ID 620 is a regular expression definition related to a date, and is a regular expression expression of "\ d {2,4} year \ d {1,2} month \ d {1,2} day" and "space deletion". Consists of regular expression parameters of. The regular expression of "\ d {2,4} year \ d {1,2} month \ d {1,2} day" is a 2-4 digit number, a "year", and a 1-2 digit number. It represents a pattern that combines a number, a "month", a one- or two-digit number, and a "day", and a character string of a date that matches this pattern can be searched. The regular expression parameter "delete space" indicates that no space character is inserted when converting an OCR string to a search string, regardless of the distance between adjacent characters.

The regular expression definition ID 630 includes a regular expression expression of "[1-9] [\ d,] * circle" and a regular expression parameter of "space = 1h". The regular expression of "[1-9] [¥ d,] * yen" starts with any number from 1 to 9, and is a pattern that combines a number including a comma with one or more digits and a "yen". , And a character string representing the amount of money that matches this pattern can be searched. The regular expression parameter "space = 1h" means that when converting an OCR character string to a search character string, the distance between adjacent characters is one or more character heights based on the character height (h). If it is free, it indicates that a space character is inserted and handled.

Using the form image 400 of FIG. 4 and the regular expression definition list 600 of FIG. 6 as examples, the image processing of the present embodiment will be described with reference to the flowchart of FIG.

First, in S501, the CPU 211 of the image forming apparatus 101 transmits the form image 400 read by the scanner 203 to the image processing server 102. The image processing server 102 acquires the form image 400 transmitted from the image forming apparatus 101.

Next, in S502, the CPU 301 of the image processing server 102 identifies the character area by performing the area analysis processing on the form image 400, and executes the character recognition processing on the character area. As a result of the character recognition process, the CPU 301 obtains the coordinates of the character area (character block), the coordinates of each character in the character area, and the character code of the character recognition result. The array of character codes for each character area obtained here is called an OCR character string. As a result of performing character recognition processing on the form image 400, it is assumed that the character strings 410 to 417 are acquired as OCR character strings.

Next, in S503, the CPU 301 of the image processing server 102 processes one of the regular expression definitions (for example, the regular expression definition 610) from the regular expression definition list 600 stored in the HDD 303.

Next, in S504, the CPU 301 of the image processing server 102 interprets the recognition result of the character string obtained in S502 based on the regular expression parameter of the regular expression definition targeted for processing in S503, and normalizes it as a search character string. do.

FIG. 7 is an example of the character recognition result. The character recognition result 701 is a character recognition result of the character string 410. The character recognition result 702 is a character recognition result of the OCR character string 411 including the item value 402. The character recognition result 703 is a character recognition result of the OCR character string 413 including the item value 403. The character line in the table of character recognition results 701 to 703 represents each recognized character, and the distance line represents the distance based on the character height as the distance to the next character. The regular expression parameter of the regular expression definition 610 is "space = 2h", which indicates that if the distance between characters is two or more character heights with respect to the character height, it is regarded as a space character. There is.

The CPU 301 scans the distance between each character and the adjacent character with respect to the character recognition result 701 of the character string 410, and detects a character that matches the space insertion condition. As a result of the detection, the character "sama" is separated from the adjacent character "contract" by a distance equivalent to 2.1 character height, so a space character is inserted here and the search character string "ABC (ABC) ( Co., Ltd. Generates an invoice.

Since the character string for searching changes for each regular expression parameter, for example, if "space = 1h" is defined, it is further between "request" and "request", and "request" and "book". Insert a space character and it will be "ABC Co., Ltd. request", and if it is defined as "space deletion", it will be "ABC Co., Ltd. invoice".

Next, the CPU 301 scans the distance between each character and the adjacent character with respect to the character recognition result 702 of the character string 411 to detect a character that matches the space insertion condition. As a result of the detection, the characters "year" and "month" are separated by two or more characters from the next character, so insert a space character here and search the character string "Issue date: May 15, 2020". Form.

The same processing is executed for the remaining character strings 412 to 417 to form a search character string for all the character strings.

Next, in S505, the CPU 301 of the image processing server 102 determines whether or not all the search character strings obtained in S504 match the regular expression expression of the regular expression definition to be processed in S503. Perform regular expression search processing.

When the regular expression expression of the regular expression definition 610 is searched for the search character string "ABC Co., Ltd. invoice" of the character string 410, the "invoice" part matches. Subsequently, when the regular expression expression of the regular expression definition 610 is searched for the search character string "issue date: May 15, 2020" of the character string 411, no matching part is obtained. Similarly, the same processing is performed for the search character strings of the remaining character strings 412 to 417 using the regular expression expression of the regular expression definition 610, and as a result, the regular expression expression is added to the other character strings. It does not match.

In S506, the CPU 301 of the image processing server 102 stores the matching information of the "invoice" obtained from the search result of S505 in the RAM 302.

Next, in S507, the CPU 301 of the image processing server 102 determines whether or not the unprocessed regular expression definition remains, and if the unprocessed regular expression definition remains, returns to S503 and returns to the unprocessed regular expression. The processing of S504 to S507 is similarly carried out with one of the definitions as the next processing target.

For example, when the regular expression definition 610 is the first processing target, the regular expression definition 620 is the next processing target. In this case, in S504, a search character string is generated for the character recognition result 702 based on the parameters of the regular expression definition 620. Since the parameter of the regular expression definition 620 is "delete space", space characters are not inserted regardless of the distance between characters. Therefore, from the character recognition result 702, "issue date: May 2020" as a search character string. 15 days "is obtained. Then, the search result of "May 15, 2020" is obtained as a part that matches the regular expression expression of the regular expression definition 620.

Similarly, when the regular expression definition 630 is processed, in S504, for the character recognition result 703, "total amount: 11,286 yen" based on the parameter "space = 1h" of the regular expression definition 630. Form a search string for. Then, "11,286 yen" is searched for in S505 as a place that matches the regular expression expression of the regular expression definition 630.

If it is determined in S507 that all the regular expression definitions have been processed, the process proceeds to S508.

In S508, the CPU 301 of the image processing server 102 executes the character string division processing based on the search result stored in the RAM in the processing of S506. The division process is a process of dividing the OCR character string at both ends of the range matched by the regular expression expression in the OCR character string. For example, in the "ABC Co., Ltd. invoice" of the OCR character string 410, the left and right sides of the "invoice" are divided as a character string delimiter. However, since the right side of the "invoice" is the right end of the OCR character string, no division occurs, and by dividing it at the position on the left side of the "invoice" (that is, between "sama" and "contract"). , The OCR string 410 is split into two OCR strings. Similarly, processing is performed for "May 15, 2020" and "11,286 yen", and the processing of the flowchart is completed.

FIG. 8 is a diagram showing the positions of the OCR character strings after applying the flowchart of FIG. 5 to the form image 400 with a rectangular frame. The OCR character string 410 is divided into OCR character strings 800 and 801, the OCR character string 411 is divided into OCR character strings 802 and 803, and the OCR character string 413 is divided into OCR character strings 804 and 805. The OCR character strings 412, 414 to 417 are unchanged.

As a result of applying this image processing, when the user names the form image 400 as a file, when the user specifies it by a click operation at the position above the document title "invoice", the character in FIG. 4 is displayed. Instead of column 410, the OCR string 801 of FIG. 8 will be selected. That is, it is possible to extract and select a character string that matches the regular expression definition.

If the search character string based on the parameters of the regular expression definition described in this embodiment is not used, the processing will be as follows. For example, when the OCR character string "ABC Co., Ltd. request" is obtained by the character recognition process of S502, the space character exists before the "book" in the regular expression expression of the regular expression definition 610. That is, the character string of "invoice" cannot be extracted. In this way, in the general character recognition process, the process of inserting space characters based on the distance between characters is applied uniformly to the entire document image, so the recognition result is as the original intention of the document creator. It is not always the case. Therefore, it is difficult to express a space with a regular expression expression.

On the other hand, in the present embodiment, since how to handle the space is defined as a parameter of the regular expression for each regular expression definition, an appropriate space character can be handled for each regular expression. For example, in a document title such as "Invoice", the character string other than the space is used as a clue for extracting the item value, and the space between the characters in the title is taken into consideration that there is a certain amount of space between characters. , "Space = 2h" and the condition of the space can be defined. Also, for example, if the format can be clearly extracted such as a date, the space can be ignored as "space deletion".

As described above, in the present embodiment, character strings often used for file names and the like are defined in advance by regular expression expressions, and at what intervals the characters of each character string are described in the document (space characters are used). Whether or not it should be included) can be defined for each character string. This makes it possible to more accurately specify the position of the character string to be divided.

If the space character output by the character recognition process is reliable, a parameter that uses the recognition result as it is may be set. For example, in the character recognition processing of an English sentence of an English document, it is assumed that a word is recognized by a word dictionary and a rutted space character is inserted between the words.

In addition, although the extraction of item values has been described in the example of file naming in this embodiment, it may be used for metadata for saving forms, automatic input of values to a document transcription system, and the like.

As described above, by applying the present embodiment, the user can divide the item value into an OCR character string suitable for file naming and the like.

<Second embodiment>
In the first embodiment, one regular expression definition is used for each item value, but the present invention is not limited to this, and a plurality of regular expression definitions may be used.

The regular expression definition list 800 in FIG. 9 is an example of a regular expression definition for a telephone number. The regular expression definition 910 is only a regular expression expression of "\ d {2,4} [\-\ (] \ d {2,4} [\-\)] \ d {4}" and has no regular expression parameter. The regular expression definition 911 is the same regular expression expression as 910, and consists of the regular expression parameter of "space deletion". Regular expression definitions The regular expression expressions of 910 and 911 are 2- to 4-digit numbers, "-" or "(", 2- to 4-digit numbers, "-" or ")", and 4-digit numbers. You can search for Japanese phone numbers that consist of.

The regular expression definition 920 is only a regular expression expression of "\ d {2,4} \ s \ d {2,4} \ s \ d {4}" and has no regular expression parameter, and the regular expression definition 921 is the same as 920. It consists of a regular expression expression and a regular expression parameter of "space = 1h". The regular expression definitions 920 and 921 can search for Japanese telephone numbers consisting of 2 to 4 digit numbers, space characters, 2 to 4 digit numbers, space characters, and 4 digit numbers. be.

FIG. 10 is an example of the character recognition result of the telephone number in the form. In the OCR character string 1010, since the space between the characters before and after the character "1" is large, a space character is inserted before and after the 1 as a result of character recognition of S502, and the character string "TEL: 0 1 23-456-7890" is inserted. It has become. In the regular expression definition 910, since there is no regular expression parameter, this OCR character string is treated as a search character string as it is, and the regular expression expression "\ d {2,4} [\-\ (] \ d {2" , 4} [\-\)] \ d {4} ”does not match. In the regular expression definition 911, since there is a parameter of "space deletion", the search character string is "TEL: 0123-456-7890", and the telephone number part can be extracted.

The OCR character string 1020 is an example in which a space character exists without a hyphen as a telephone number delimiter. Since the space between the characters before and after the character "1" in the OCR character string 1020 is large, a space character is inserted before and after the 1 as a result of character recognition in S502, resulting in the character string "TEL: 0 1 23 456 7890". There is. In the regular expression definition 920, since there is no regular expression parameter, this OCR character string is treated as a search character string as it is, and the regular expression expression "\ d {2,4} \ s \ d {2,4} \ Searching for "s \ d {4}" does not match. In the regular expression definition 921, since there is a parameter of "space = 1h", the search character string is "TEL: 0123-456-7890", and the telephone number part can be extracted.

When trying to match such a space character output as a character recognition result only with a regular expression expression, for the regular expression expression "\ d {2,4}" of the first 2 to 4 digits. There is no choice but to write "[\ d \ s] {2,7}". That is, since the number of spaces is indefinite, the definition of the number of digits is also defined as 2 to 7 digits assuming the maximum space, but in such a regular expression, many character strings match. This will increase noise other than the item values that you originally want to extract.

As described above, by applying this embodiment, it is possible to reduce the influence of the space character output as the character recognition result and simply define the regular expression expression.

<Third embodiment>
In the first embodiment, the pattern to be extracted is defined by the regular expression definition and divided at the matching left and right ends, but the operation other than the division may be specified by the regular expression parameter.

The regular expression definition list 1100 of FIG. 11 defines the operation when the regular expression expression is matched as the regular expression parameter 2. In the regular expression definition 1111, the regular expression expression is ". * No", for example, it represents a combination of a sequence of any characters and "no", the regular expression parameter is "space deletion", and the division is defined by the action definition. is doing. The division is the operation of dividing at the left and right ends of the search result described in the first embodiment. The regular expression definition 1112 represents a character string of "nov", the regular expression parameter is "space deletion", and protection is defined in the action definition. Protect is a definition of the operation that does not divide the range that matches the search result.

The character string 1201 “Invoice No 10123” in FIG. 12 represents an invoice number, and when only the regular expression definition 1111 is used as the regular expression definition list, as a result of the processing of the flowchart of FIG. 5, it becomes “Invoice No”. Match. As a result, it is divided into "Invoice No" and "10123".

The character string 1202 “12 Nov 2019” in FIG. 12 represents a date, and when only the regular expression definition 1111 is used in the regular expression definition list, the result of the search matches “12 No”. As a result, it is divided into "12 No" and "v 2019", which is not preferable for the user who wants to extract the date. By searching by adding the regular expression definition 1112 to the regular expression definition list, the search result that matches "Nov" of the character string 1202 and does not perform division within that range is stored by S506 and divided by S508. Information is protected, division is not performed, and unintended division can be prevented.

In addition, other operations may be specified. In the regular expression definition 1120, as a submatch division, it is possible to define that the grouping of regular expression expressions is used to divide the match on the left and right of the submatch. As a result of searching the regular expression "(. * No) (\ d *)" of the regular expression definition 1120 for the character string 1201, "Invoice No" is matched in submatch 1 and "10123" is matched in submatch 2. , In the meantime, carry out the division.

As described above, by applying the present embodiment, it is possible to prevent unintended division or to carry out more complicated division.

<Fourth Embodiment>
In the first embodiment, the method of extracting the item value is realized by dividing the OCR character string, but it may be combined. For example, in the case where the OCR character string 411 of the form image 400 is divided into three OCR character strings of "issue date: 2020" and "May" and "15th" as a result of character recognition, S503 In the search character string generation process of, after combining the OCR character strings in the vicinity of the same character height, the search is performed by the regular expression expression of S504, and if they match, the matching range is combined.

As described above, by applying the present embodiment, it is possible to extract by a regular expression expression without being affected by the particle size of the OCR character string output as the character recognition result.

<Other embodiments>
INDUSTRIAL APPLICABILITY The present invention supplies a program that realizes one or more functions of the above-described embodiment to a system or device via a network or storage medium, and one or more processors in the computer of the system or device reads and executes the program. It can also be realized by the processing to be performed. It can also be realized by a circuit (for example, ASIC) that realizes one or more functions.

Although the preferred embodiments of the present invention have been described above, the present invention is not limited to these embodiments, and various modifications and modifications can be made within the scope of the gist thereof.

100 Image processing system 101 Image forming device 102 Image processing server (image processing device)
103 User terminal 104 Network

Claims (7)

A storage means for storing a regular expression definition in which a character string to be extracted is associated with a regular expression expression and a parameter related to a space character defined in advance for each regular expression expression.
A character recognition means that executes character recognition processing on a document image,
A generation means for generating a search character string by inserting a space character into the character string of the character recognition result by the character recognition means based on the parameters related to the space character.
A search means for searching a part matching the regular expression expression from the search character string generated by the generation means, and a search means.
A division means for dividing the character string of the character recognition result based on the portion searched by the search means, and a division means.
An image processing apparatus characterized by having. The generation means inserts the space character at a position where the distance between adjacent characters in the character string of the character recognition result is larger than the interval defined as a parameter for the space character, thereby producing the search character string. The image processing apparatus according to claim 1, wherein the image processing apparatus is generated. The generation means has the character recognition result for the character height in the character string of the character recognition result, as opposed to the regular expression definition in which the parameter defined by using the character height is defined as the parameter for the space character. The second aspect of claim 2, wherein the search character string is generated by inserting the space character at a position where the distance between adjacent characters of the character string is larger than the parameter related to the space character. Image processing device. For the regular expression definition in which the space is deleted as a parameter related to the space character, the generation means does not insert the space character in the character string of the character recognition result, and the search character is used. The image processing apparatus according to any one of claims 1 to 3, wherein a column is generated. When the position specified by the user in the document image is a position corresponding to the character string divided by the dividing means, the document image is further provided with a selection means for selecting the character string after the division. The image processing apparatus according to any one of claims 1 to 4. It is an image processing method that divides the character string of the character recognition result based on the regular expression definition that associates the character string to be extracted with the regular expression expression and the parameter related to the space character defined in advance for each regular expression expression. hand,
A character recognition step that executes character recognition processing on a document image,
A generation step of generating a search character string by inserting a space character into the character string of the character recognition result obtained in the character recognition step based on the parameters related to the space character.
A search step for searching a part matching the regular expression expression from the search character string generated in the generation step, and a search step.
A division step for dividing the character string of the character recognition result based on the location searched in the search step, and a division step.
An image processing method characterized by having. A program for making a computer function as each means of the image processing apparatus according to any one of claims 1 to 5.

Images