JP7111143B2 - Image processing device, image processing method and program - Google Patents

Description

The present invention relates to an image processing device, an image processing method, and a program.

In relation to the form format, Patent Document 1 describes a method for efficiently selecting the most suitable form format for the form. In this method, the form formats are grouped and one representative form format is defined for each group. In this method, any one group is selected based on the feature match rate between the read form image and the representative form format. Furthermore, in this method, the form format having the highest feature match rate with the read form image is selected from among the form formats in the selected group.

In addition, in relation to correction of character recognition results, Patent Document 2 discloses that when unreadable characters (reject characters) occur in character recognition when multiple types of forms are mixed, correction of unreadable characters is performed. It is described that by designating the type of form to be processed and correcting unreadable characters in the designated type of form, the same type of form is continuously corrected.

In addition, in relation to correction of character recognition results, Patent Document 3 discloses that when correcting/confirming an image-inputted (character-recognized) form, input contents are displayed according to a predetermined format and corrections are accepted. and that when a correction item is instructed by the user, a specific image portion corresponding to the received correction item is highlighted.

JP 2016-048444 A Japanese Patent Application Laid-Open No. 2004-118380 JP-A-2002-007951

When confirming and correcting character recognition results, it is important that the confirmation/correction operator be able to ascertain which item is shown where. In particular, when a worker confirms and corrects the result of reading a document with an unfamiliar format, it may take time to grasp where the item to be confirmed is written.
Therefore, it is preferable to assist the operator in understanding which item is displayed where. At that time, it is preferable to be able to assist even if the format of the document to be subjected to character recognition is not known in advance.

An object of the present invention is to provide an image processing apparatus, an image processing method, and a program that can solve the above-described problems.

According to the first aspect of the present invention, an image processing apparatus performs character string recognition for recognizing a character string of a specific item in a recognition target image based on learning results using a plurality of images including character strings. and the image to be recognized and the result of character string recognition by the character string recognition unit in such a manner that a correspondence relationship indicating the same specific item in the image to be recognized and the character string recognition result by the character string recognition unit can be understood. and an output unit for outputting , wherein the character string recognition unit identifies the character string candidates of the specific item in the recognition target image, and the output unit outputs a range of the identified character string candidates. is output in a comprehensible manner .

According to a second aspect of the present invention, an image processing method includes a step of recognizing a character string of a specific item in an image to be recognized based on a result of learning using a plurality of images including character strings; a step of outputting the recognition target image and the character string recognition result in a manner in which a correspondence relationship indicating the same specific item in the recognition target image and the character string recognition result can be grasped ; In the step of recognizing the character string, the candidate character string of the specific item in the image to be recognized is specified, and in the step of outputting, the range of the specified candidate character string is grasped. output .

According to a third aspect of the present invention, a program instructs a computer to recognize a character string of a specific item in an image to be recognized based on learning results using a plurality of images containing character strings; and outputting the recognition target image and the character string recognition result in such a manner that a correspondence relationship indicating the same specific item in the recognition target image and the character string recognition result can be grasped. and, in the step of recognizing the character string, specifying the candidate character string of the specific item in the image to be recognized, and in the step of outputting, a range of the specified candidate character string can be grasped. This is a program for outputting with .

According to the present invention, even if the format of the document to be subjected to character recognition is not known in advance, it is possible to assist the operator in recognizing which item is shown where.

1 is a diagram showing a device configuration example of an image processing system including an image processing device according to an embodiment; FIG. It is a figure which shows the hardware structural example of the image processing apparatus which concerns on embodiment. 1 is a schematic block diagram showing the functional configuration of an image processing apparatus according to a first embodiment; FIG. FIG. 4 is a diagram showing an example of a document form; It is a figure which shows the outline|summary of the recording table which the database based on 1st embodiment memorize|stores. FIG. 3 is a first diagram showing the processing flow of the image processing apparatus according to the first embodiment; FIG. 7 is a second diagram showing the processing flow of the image processing apparatus according to the first embodiment; FIG. 8 is a diagram showing an example of a display screen in which a display of recorded character strings and an image of a document form are displayed side by side by the display unit according to the first embodiment; FIG. 10 is a diagram showing an example of a display screen on which the display unit according to the first embodiment displays a correspondence relationship between a display of recorded character strings and an image of a document form; FIG. 6 is a schematic block diagram showing the functional configuration of an image processing apparatus according to a second embodiment; FIG. 11 is a first diagram showing a processing flow of an image processing device according to a second embodiment; FIG. 9 is a second diagram showing the processing flow of the image processing apparatus according to the second embodiment; 1 is a diagram illustrating an example of the configuration of an image processing apparatus according to an embodiment; FIG.

Embodiments of the present invention will be described below, but the following embodiments do not limit the invention according to the claims. Also, not all combinations of features described in the embodiments are essential for the solution of the invention.

FIG. 1 is a diagram illustrating an example configuration of an image processing system including an image processing apparatus according to an embodiment.
With the configuration shown in FIG. 1, the image processing system 100 comprises an image processing device 1, an image reading device 2, a recording device 3, and a database 4. FIG.
The image processing device 1 is connected to the image reading device 2 by a communication cable. The image reading device 2 optically acquires image data such as a document form and outputs the image data to the image processing device 1 . The image processing apparatus 1 performs OCR processing on the image data of the document form and recognizes the characters. The image processing device 1 outputs the character recognition result to the recording device 3, and the recording device 3 records the character recognition result in a database. Note that the document targeted by the image processing apparatus 1 is not limited to a specific type. Various OCR-processable documents can be processed by the image processing apparatus 1 .

A database 4 is connected to the image processing device 1 and the recording device 3 . The database 4 stores correspondence relationships between image data of a plurality of document forms registered in the past from the recording device 3 and recorded character strings indicating character strings to be recorded among the character strings included in the image data. The character string indicated by the recorded character string is an important character string that should be recorded and saved in the database 4 among the character strings described in the document form. An operator who uses the image processing system 100 registers image data of a plurality of document forms registered in the past using the recording device 3 and recorded character strings among the character strings included in the image data in the database 4 in advance. Keep
The operator is also called a user of the image processing apparatus 1 or simply a user.

In the database 4, the correspondence relationship between the image data of the document form and the information of the record character string indicating the character string to be recorded among the character string information included in the image data is sufficiently recorded for many document forms. It shall be In such a state, the image processing apparatus 1 performs processing.

FIG. 2 is a diagram showing a hardware configuration example of the image processing apparatus.
With the configuration shown in FIG. 2 , the image processing apparatus 1 includes an image processing apparatus body 10 , a display device 17 and an input device 18 .
The image processing apparatus main body 10 includes a CPU (Central Processing Unit) 11, an IF (Interface) 12, a communication module 13, a ROM (Read Only Memory) 14, a RAM (Random Access Memory) 15, an HDD (Hard Disk Drive) 16, and the like. A computer with a configuration. The communication module 13 may perform wireless communication with the image reading device 2, the recording device 3, and the database 4, or may perform wired communication. good.
The display device 17 has a display screen such as a liquid crystal panel or an LED (Light Emitting Diode) panel.
The input device 18 is a device that receives user operations, such as a keyboard and mouse, a touch sensor that is provided on the display screen of the display device 17 and constitutes a touch panel, or a combination thereof.

<First embodiment>
FIG. 3 is a schematic block diagram showing the functional configuration of the image processing apparatus 1 according to the first embodiment.
The communication unit 110 is configured using the communication module shown in FIG. 2, and communicates with other devices. In particular, the communication unit 110 communicates with the image reading device 2, the recording device 3, and the database 4. FIG.
The display unit 120 is configured using the display device 17 in FIG. 2, and displays various images. In particular, the display unit 120 corresponds to an example of an output unit, and the first document image and the second document image displayed corresponding to the first document image correspond to the same specific item. to output

For example, the display unit 120 displays an image of a document form (a raw image, not an OCR result) as a first document image, and a recorded character string that is a processing result of the image processing apparatus 1 as a second document image. A GUI screen image displayed in a predetermined format is displayed. Then, the display unit 120 indicates which character string in the first document image the character string indicated in the second document image is the reading result of, by drawing a line between both character strings.

Alternatively, when the correct character string of the recorded character string is obtained, the display unit 120 displays the OCR result of the document form as the first document image in the format of the document form, and displays the OCR result of the document form as the second document image. A GUI screen image displaying the correct character string in a predetermined format may be displayed. Then, the display unit 120 may indicate which character string in the first document image the character string indicated in the second document image indicates the correct character string by drawing a line between the two character strings. good.
However, the method by which the output unit outputs the image and related information is not limited to the method of displaying these images and related information. For example, the communication unit 110 may function as an output unit and transmit images and related information to another device for display.

The operation input unit 130 is configured using the input device shown in FIG. 2 and receives user operations. In particular, operation input unit 130 receives correction operations for character strings displayed by display unit 120 .
The storage unit 180 is configured using the ROM 14, RAM 15 and HDD 16 of FIG. 2, and stores various data.
The control unit 190 is configured by the CPU 11 in FIG. 2 reading and executing a program from the storage unit 180 (ROM 14, RAM 15 and HDD 16 in FIG. 2). The control unit 190 controls each unit of the image processing apparatus 1 to execute various processes.
The acquisition unit 191 acquires image data of a document form.

A feature amount extraction unit 192 extracts a feature amount of a character string from the recognition processing result of the image data of the document form. For example, the feature amount extracting unit 192 extracts a first feature amount indicating the feature of the recorded character string included in the image data of the document form for each image data of the document form, based on the recognition processing results of the image data of a plurality of document forms. Extract. Specifically, the feature amount extraction unit 192 identifies a recorded character string included in the document image that has undergone character recognition processing, and extracts a first feature amount of the identified recorded character string in the document image. The identification of the recorded character string here means determining any one of the character strings in the document form as one recorded character string. Extraction of the feature quantity is also referred to as generation of the feature quantity.

The recording unit 193 extracts and records the recorded character string from the character string information read from the image data of the new document form using the feature amount of the character string in the image data of the new document form. In particular, the recording unit 193 corresponds to an example of a character string detection unit, and is recorded in advance based on the result of learning using a plurality of document images, and for each type of document image and for each specific item, the character string of the item is detected. Character strings of specific items in the first document image are detected based on the feature amount of the displayed first document image (document form image) among the feature amounts (first feature amount) indicating the features. do.

Also, the recording unit 193 detects a character string of the same item as the specific item in the first document image from the second document image. When the format of the second document image is known, the recording unit 193 uses the format information to detect the recorded character string in the second document image. On the other hand, when the recording unit 193 does not know the format of the second document image, the feature amount extraction unit 192 extracts the first feature amount also for the second document image, and the recording unit 193 stores the obtained first feature Quantities may also be used to specify record strings.

Also, the combination of the feature amount extracting unit 192 and the recording unit 193 corresponds to an example of a correspondence learning unit, and machine-learns the correspondence of the same specific item between the first document image and the second document image. .
For example, when the recording unit 193 knows the format of the second document image, the feature amount extraction unit 192 extracts the feature amount of the character string in the first document image (image of the document form) by machine learning, and records it. A unit 193 identifies the recorded character string in the first document image based on the obtained character string feature amount. As a result, the recording unit 193 knows the recorded character strings for both the first document image and the second document image, and acquires the correspondence relationship between the recorded character strings between these two document images. .

Alternatively, if the recording unit 193 does not know the format of the second document image, the feature amount extraction unit 192 extracts the character string feature amount for the second document image in addition to the first document image. may In this case, the recording unit 193 specifies the recorded character string based on the obtained character string feature amount for each of the first document image and the second document image. As a result, the recording unit 193 knows the recorded character strings for both the first document image and the second document image, and acquires the correspondence relationship between the recorded character strings between these two document images. .

The correspondence learning unit (combination of the feature amount extraction unit 192 and the recording unit 193) uses the image after the character string of the second document image is determined to determine the first document image and the second document. Machine learning may be performed on the correspondence between the same specific item and the image. If the character string in the second document image is confirmed, the confirmed character string is considered correct. By performing machine learning using this accurate character string, the correspondence learning unit can learn the correspondence of the same specific item between the first document image and the second document image with relatively high accuracy. is expected.

Through such processing, the image processing apparatus 1 reduces labor for recording character string information to be recorded contained in image data of a new document form.

FIG. 4 is a diagram showing an example of a document form.
As shown in this figure, in the document form, the mark of the company that created the document, the date of creation, the person in charge of creation, and the content of the document are described in a format unique to the document form. For example, if the document form is an order form, one or more sets of information such as the name of the ordered product and the number of ordered products are indicated. Based on one document form, the operator uses the recording device 3 to record a specific character string (recording character string) to be recorded among the character strings described in the document form in the database 4 . Specifically, the operator inputs a recording character string to be recorded in the database 4 by the recording device 3 while looking at the document form. Also, the operator causes the image reading device 2 to read the image data of the document form. The document form is read by the image reading device 2 and output to the image processing device 1 based on the operator's operation. Based on the operation of the operator and the control of the image processing apparatus 1, the recording device 3 associates the image data of one document form with the recorded character string among the character strings described in the document form. Record in database 4. In the example of FIG. 4, the date 51, orderer 52, product name 53, quantity 54, and amount 55 are recorded character strings. Other information such as non-recorded character strings that are not recorded by the operator are also printed on the document form 5 . The information includes, for example, the name 501 of the orderer who issued the document, the emblem image 502 of the orderer, the title 503 of the document, the greeting 504, and the like.

FIG. 5 is a diagram showing an outline of a record table stored in the database.
As shown in FIG. 5, the database 4 associates the image data of the document form with the recorded character strings among the character strings described in the document form and stores them in a record table.

FIG. 6 is a first diagram showing the processing flow of the image processing apparatus according to the first embodiment. FIG. 6 shows an example of a processing procedure for the image processing device 1 to extract the first feature amount.
Next, the processing flow of the image processing apparatus 1 will be explained step by step.
First, in the database 4, a combination of image data for a certain document form and a record character string described in the document form is recorded for a plurality of document forms of the same format. For example, it is assumed that the recorded character string information (information indicating the recorded character string) in the format of the document form 5 shown in FIG. 4 is recorded for a plurality of sheets.
As a combination of these image data and recorded character string information, for example, image data and recorded character string information of document forms handled in past business can be used. If the necessary amount of image data and recorded character string information can be secured from past work, there is no need to separately prepare the image data and recorded character string information in order to cause the image processing apparatus to acquire the first feature amount.
In such a state, the operator activates the image processing apparatus 1 and instructs the image processing apparatus 1 to start processing.

The acquisition unit 191 of the image processing apparatus 1 controls the communication unit 110 to read the image data of the document form and the information of the recorded character string corresponding to the image data from the database 4 (step S601). Acquisition unit 191 outputs the image data and the recorded character string to feature amount extraction unit 192 . The feature amount extraction unit 192 performs OCR processing on the image data to detect all character strings in the image data and coordinates in the image data indicating the range of the character strings (step S602). A character string is a group of characters composed of a plurality of characters. The feature amount extracting unit 192 analyzes the range of one group based on the interval with other characters, extracts one or more characters included in the range as a character string, and extracts the character string in the image data. Find the coordinates that indicate the extent of . Characters included as character strings may include symbols such as ideograms and phonetic characters, marks, icon images, and the like.

The feature quantity extraction unit 192 compares the character string extracted from the image data by OCR processing with the recorded character string read from the database 4 together with the image data. The feature amount extracting unit 192 extracts character strings in the image data that match character information of the recorded character strings among the character strings extracted from the image data by OCR processing, attributes of the characters included in the character strings, and their ranges. coordinates are specified (step S603).

The attributes of characters (character attributes) here are information represented by numbers, alphabets, hiragana, kanji, number of characters, character height, font, and the like. The coordinates of the character string range are coordinates indicating the position of the character string in the document form. For example, the coordinates of the character string range may be information indicating the coordinates of the leading character and the coordinates of the ending character included in the character string. Hereinafter, attributes of characters included in a character string and coordinates within the range of the character string are collectively referred to as attributes of the character string or character string attributes.

Also, the character information here may be only a character string, or may include a character string attribute. That is, the feature amount extraction unit 192 may determine whether or not the recorded character string and the character string in the image data are the same as character strings. Alternatively, the feature quantity extraction unit 192 may determine the identity of character string attributes in addition to the identity of characters.

If the feature amount extraction unit 192 cannot uniquely identify a character string whose character information matches the recorded character string, the image processing apparatus 1 excludes the document image from the processing target (first feature amount extraction target). You may do so. Alternatively, the image processing apparatus 1 may cause the display unit 120 to display an image in which the range of each recorded character string candidate is framed, and specify the recorded character string in the character string selected by the operator. . The recorded character string candidate here is a character string that is associated with a recorded character string that is determined not to be uniquely identified among character strings whose character information matches the character information of the recorded character string. . Further, specifying the recorded character string here means determining any one of the character strings in the document form as one recorded character string.
When the feature quantity extraction unit 192 determines that the character information of each of a plurality of character strings in the document form matches the character information of one recorded character string, these character strings are candidates for the recorded information. Become. The recorded character string is uniquely specified by the operator selecting any one of the plurality of character strings.

Next, the feature quantity extraction unit 192 extracts a feature quantity common to document forms of the same format and for each recorded character string using the character string attributes extracted for each document form and for each recorded character string (step S604). .
Specifically, for each recorded character string, the feature quantity extraction unit 192 analyzes the character string attributes of the recorded character string in a plurality of document forms, and extracts one characteristic quantity for each recorded character string.
The method by which the feature amount extraction unit 192 extracts the feature amount for each recorded character string that is common to document forms of the same format is not limited to a specific method. For example, for a plurality of character string attributes obtained from a plurality of document forms, the feature amount extracting unit 192 may determine the coordinates of the leading character, the coordinates of the trailing character, the type of character, the height of the character, the type of font, and the like. A mode may be obtained for each item. In addition, the feature quantity extraction unit 192 calculates the average value (Average) or the median value (Median ) may be obtained. In addition, the feature quantity extraction unit 192 may use a feature quantity having a range, such as using the maximum value and the minimum value as a feature quantity for an item represented by a numerical value, or a feature quantity represented by a plurality of numerical values. can be Further, the feature quantity extraction unit 192 may obtain the feature quantity by digitizing non-numerical attributes such as the type of character and the type of font. Also, the feature quantity extraction unit 192 may extract the feature quantity using a known machine learning algorithm.
When the feature amount extraction unit 192 acquires a plurality of numerical values for one format and one recorded character string of a document form, the plurality of numerical values may be vectorized to extract a feature amount of one vector. .

A feature amount common to document forms of the same format and for each recorded character string extracted by the feature amount extraction unit 192 is referred to as a first feature amount. The feature quantity extraction unit 192 uses a plurality of document forms of the same format and extracts the first feature quantity of each recorded character string in the format. The first feature quantity is a feature quantity for extracting a recorded character string. The first feature amount may include information indicating character attributes, coordinates indicating the range of character strings, or a combination thereof.
The feature quantity extraction unit 192 records the first feature quantity obtained for each recorded character string in the database 4 in association with the format identifier of the document form (step S605).

For example, the feature amount extraction unit 192 extracts the character attributes and character string ranges of the date 51, orderer 52, product name 53, quantity 54, and amount 55, which are recorded character strings included in the format of the document form 5 in FIG. Each first feature quantity indicating coordinates indicating the , etc. is recorded in the database 4 in association with the format identifier of the document form 5 .
After step S605, the image processing apparatus 1 ends the processing of FIG.

Through the above processing, the image processing apparatus 1 can extract information (first feature amount) used for reducing the labor of the operator to record the record character string, and store the extracted information in the database 4 . As a result, the image processing apparatus 1 can receive input of image data of a new document form and automatically record the record character string included in the new document form in the database 4 . The processing will be described with reference to FIG.

FIG. 7 is a second diagram showing the processing flow of the image processing apparatus according to the first embodiment. FIG. 7 shows an example of a processing procedure in which the image processing apparatus 1 extracts a recorded character string from newly input image data.
The operator performs an operation to cause the image reading device 2 to read a new document form. As a result, the image reading device 2 generates image data of the document form and outputs (transmits) it to the image processing device 1 . The acquisition unit 191 of the image processing apparatus 1 acquires image data from the data received by the communication unit 110 (step S701). Acquisition unit 191 outputs the image data to feature amount extraction unit 192 . The feature amount extraction unit 192 performs OCR processing on the image data, and extracts a character string, characteristics of characters (character attributes) included in the character string, and coordinates in the image data within the range of the character string for each character string. Detect (step S702). The feature amount extraction unit 192 extracts the third feature amount obtained by converting the detected information into the feature amount for each character string in the image data (step S703). That is, the third feature amount is information indicating the feature of the character string included in the document form of the newly read image data. After that, the feature quantity extraction unit 192 reads out the first feature quantity for each recorded character string from the database 4 (step S704). The feature amount extraction unit 192 outputs the third feature amount and the first feature amount to the recording unit 193 .

The recording unit 193 acquires the third feature amount for each character string in the image data and the first feature amount for each recorded character string, and associates the first feature amount and the third feature amount (step S705). Specifically, the recording unit 193 associates each of the first feature amounts with a third feature amount that matches the feature amount or one of the closest third feature amounts. With this association, a recorded character string is selected from the character strings obtained by OCR processing the image data of the document form.
Next, the feature amount extraction unit 192 controls the display unit 120 to display the recorded character string, which is the result of processing by the image processing apparatus 1, and the image of the document form side by side (step S706).

FIG. 8 is a diagram showing an example of a display screen on which the display unit 120 displays a recorded character string and an image of a document form side by side. In the example of FIG. 8, the display unit 120 displays a document form image W1 and a recorded character string display window W2. Character strings resulting from OCR of the character strings C1a and C1b in the document image W1 are displayed as character strings C2a and C2b, which are recorded character strings, in the recorded character string display window W2. The "OK" button icon B1 is a button icon that the operator presses to confirm and correct the recorded character string displayed in the recorded character string display window W2. The pressing operation of the button icon is performed, for example, by a touch operation on the button icon, a mouse click, or the like.

Next, the recording unit 193 determines whether the correspondence relationship between the display of the recorded character string and the image of the document form has been acquired (step S707). For example, the recording unit 193 determines that the correspondence has been acquired when the number of times of learning for grasping the correspondence between the display of the recorded character string and the image of the document form is equal to or greater than a predetermined number. As described above, the learning for grasping the correspondence relationship between the display of the recorded character string and the image of the document form may be the learning for specifying the recorded character string in the display of the recorded character string, or the learning for identifying the recorded character string in the display of the recorded character string. It may be learning for identifying the recorded character string in the image of the form. Alternatively, it may be a combination of these learnings.

If it is determined that the correspondence relationship between the display of the recorded character string and the image of the document form has been acquired (step S707: YES), the recording unit 193 controls the display unit 120 to display the recorded character string and the image of the document form. The correspondence with the image is displayed (step S708).
FIG. 9 is a diagram showing an example of a display screen on which the display unit 120 displays the correspondence relationship between the display of the recorded character string and the image of the document form.
FIG. 9 shows an example of displaying the correspondence relationship between the display of the recorded character string and the image of the document form on the display screen of FIG. Character strings C1a, C1b, C2a, and C2b are shown surrounded by rectangles F1a, F1b, F2a, and F2b, respectively. A line La indicates that the character strings C1a and C2a correspond to each other. A line Lb indicates that the character strings C1b and C2b correspond.
In this manner, the display unit 120 displays the correspondence relationship between the display of the recorded character string and the image of the document form, so that the operator can select either the format of the document form or the format of the display window of the recorded character string. Even if you are unfamiliar with both, you can easily and reliably grasp the correspondence between character strings.

When the operator confirms and corrects the recorded character string and then performs a confirmation operation, the image processing apparatus 1 acquires the corrected recorded character string (step S709). If the recording unit 193 determines in step S707 that the correspondence relationship between the display of the recorded character string and the image of the document form has not been acquired (step S707: NO), the process also proceeds to step S709.
Then, the feature quantity extraction unit 192 and the recording unit 193 as the correspondence relationship learning unit learn the coordinates of the recorded character string in the display of the processing result (in the example of FIGS. 8 and 9, the recorded character string display window W2). step S710). For example, the feature amount extraction unit 192 extracts the first feature amount for display of the processing result in the same manner as in the document form, and the recording unit 193 uses the first feature amount to determine the position of the recorded character string. Identify and remember.
If the recording unit 193 already knows the display format of the processing result, the processing of step S710 is unnecessary.

The recording unit 193 records the recording character string in the recording table in association with the identification information of the document form (step S711).
For example, assume that a third feature amount a3, a third feature amount b3, a third feature amount c3, and a third feature amount d3 have been acquired from the image data of the document form. Then, the third feature amount a3 is the first feature amount a1 recorded in advance in the database, the third feature amount b3 is the first feature amount b1, the third feature amount c3 is the first feature amount c1, and the third feature amount It is assumed that the feature amount of the quantity d3 matches the first feature amount d1. In this case, the recording unit 193 records the character strings corresponding to the third feature amount a3, the third feature amount b3, the third feature amount c3, and the third feature amount d3 in the record table of the document form as recorded character strings. do. The character string corresponding to the third feature amount referred to here is the character string from which the third feature amount is extracted. When the operator corrects the recorded character string, the recording unit 193 records the corrected recorded character string in the recording table of the document form.
After step S711, the image processing apparatus 1 ends the processing of FIG.

The image processing apparatus 1 may update the first feature amount in the document form in the process of FIG. For example, in step S710, the image processing apparatus 1 may learn the first feature amount in the document form in addition to or instead of the machine learning of the first feature amount in the processing result display. In that case, the image processing apparatus 1 may perform the processing flow of FIG. 6 again. Alternatively, the image processing apparatus 1 may perform additional learning in step S710 to update the first feature amount without reprocessing the data that has already been processed in FIG.
The image processing device 1 updates the first feature quantity by the processing of FIG. 7, thereby increasing the number of sample data and improving the accuracy of the first feature quantity, thereby increasing the accuracy with which the image processing device 1 extracts the recorded character string. is expected to improve. In addition, when a recorded character string is added in the process of FIG. 7, the image processing apparatus 1 can extract the newly added recorded character string from the image data, thus saving the operator the trouble of inputting the character string. It is expected that it will be saved.

According to the processing shown in FIG. 7, the image processing apparatus 1 converts the recorded character string in the image data of the newly input document form based on the image data of the document form recorded in advance by the operator and the recorded character string. Can be automatically recorded. Therefore, the image processing apparatus 1 can reduce the labor of the operator for recording the record character string in the document form. Also, the operator can refer to the display on the display unit 120 to check and correct the recorded character string that is the processing result of the image processing apparatus 1 . At this time, the display unit 120 indicates the correspondence between the character strings in the display of the processing result and the character strings in the image of the document form, so that the operator can easily understand the correspondence.

<Second embodiment>
In the second embodiment, a case where the image processing apparatus 1 supports a plurality of formats of document forms will be described.
FIG. 10 is a schematic block diagram showing the functional configuration of an image processing apparatus according to the second embodiment.
As shown in FIG. 10, the image processing apparatus 1 according to the second embodiment has functions of a group classifying section 194 and a group identifying section 195 in addition to the functional sections shown in FIG.

FIG. 11 is a first diagram showing the processing flow of the image processing apparatus according to the second embodiment.
Next, the processing flow of the image processing apparatus 1 according to the second embodiment will be explained step by step.
In the database 4, a large number of combinations of image data for a plurality of document forms with different formats and recorded character strings described in each document form are recorded for each document form. In such a state, the operator activates the image processing apparatus 1 and instructs the image processing apparatus 1 to start processing.

The acquisition unit 191 of the image processing apparatus 1 determines whether all the image data of the document form and the information of the recorded character string corresponding to the image data have been read from the database 4 (step S901). In the case of NO, the acquisition unit 191 reads the image data of the document form and the information of the recorded character string corresponding to the image data from the database 4 (step S902). Acquisition unit 191 outputs the image data and the recorded character string to feature amount extraction unit 192 . The feature amount extraction unit 192 performs OCR processing on the image data to detect all character strings in the image data and coordinates in the image data (step S903). A character string is a group of characters composed of a plurality of characters. The feature amount extracting unit 192 analyzes the range of one group based on the interval with other characters, extracts one or more characters included in the range as a character string, and extracts the character string in the image data. Find the coordinates that indicate the extent of . Characters included as character strings may include symbols such as ideograms and phonetic characters, marks, icon images, and the like.

The feature quantity extraction unit 192 compares the character string extracted from the image data by OCR processing with the recorded character string read from the database 4 together with the image data. The feature quantity extracting unit 192 extracts character strings in the image data that match character information of the recorded character strings among the character strings extracted from the image data by OCR processing, attributes of the characters included in the character strings, and their ranges. coordinates are identified (step S904).

As described in the first embodiment, character attributes are information represented by numbers, alphabets, hiragana, kanji, number of characters, character height, font, and the like. The coordinates of the character string range are coordinates indicating the position of the character string in the document form. For example, the coordinates of the character string range may be information indicating the coordinates of the leading character and the coordinates of the ending character included in the character string. Attributes of characters included in a character string and coordinates within the range of the character string are collectively referred to as character string attributes or character string attributes.

As in the case of the first embodiment, the character information here may be only character strings, or may include character string attributes. That is, the feature amount extraction unit 192 may determine whether or not the recorded character string and the character string in the image data are the same as character strings. Alternatively, the feature quantity extraction unit 192 may determine the identity of character string attributes in addition to the identity of characters.

If the feature amount extraction unit 192 cannot uniquely identify a character string whose character information matches the recorded character string, the image processing apparatus 1 excludes the document image from the processing target (first feature amount extraction target). You may do so. Alternatively, the image processing apparatus 1 may cause the display unit 120 to display an image in which the range of each recorded character string candidate is framed, and specify the recorded character string in the character string selected by the operator. . As described in the first embodiment, the recorded character string candidate here is a recorded character string that is determined not to be uniquely identified among character strings whose character information matches the character information of the recorded character string. is a string associated with Further, specifying the recorded character string here means determining any one of the character strings in the document form as one recorded character string.
When the feature quantity extraction unit 192 determines that the character information of each of a plurality of character strings in the document form matches the character information of one recorded character string, these character strings are candidates for the recorded information. Become. The recorded character string is uniquely specified by the operator selecting any one of the plurality of character strings.

Next, the feature quantity extraction unit 192 extracts a feature quantity for each document form and each recorded character string using the character string attributes extracted for each document form and each recorded character string (step S905). Specifically, the feature quantity extraction unit 192 converts the character string attribute of the character string associated with the recorded character string in step S904 into a feature quantity. Since a plurality of types of formats are targeted in the second embodiment, at the time of step S905 when document forms are not grouped by format, unlike the case of step S604 in FIG. 7, the first feature value is directly extracted. can't. Therefore, the feature quantity extraction unit 192 extracts the feature quantity for each document form and for each recorded character string in preparation for extracting the first feature quantity for each group. The feature amount for each document form and for each recorded character string is called an individual first feature amount.
The feature amount extraction unit 192 records the obtained individual first feature amount in the database 4 in association with the identifier of the document form and the identifier of the recording character string (step S906). As the identifier of the recorded character string, for example, a coordinate value indicating the position of the recorded character string can be used.

For example, the feature amount extraction unit 192 extracts the character attributes and character string ranges of the date 51, orderer 52, product name 53, quantity 54, and amount 55, which are recorded character strings included in the format of the document form 5 in FIG. is recorded in the database 4 in association with the identifier of the document form 5 and the identifier of the recorded character string for each document form and each recorded character string.

The feature amount extraction unit 192 also extracts non-recorded character strings in the image data that do not match the character information included in the recorded character strings and character string attributes of the non-recorded character strings (step S907). As described above, the non-recorded character string is a character string that is not recorded by the operator, that is, a character string other than the recorded character string. The character string attribute may include either or both of information indicating the attributes of characters included in the character string and information indicating the coordinates of the range of the character string.

The feature amount extraction unit 192 extracts the feature amount for each document form and each non-recorded character string using the character string attributes extracted for each document form and each non-recorded character string (step S908).
Specifically, for each character string that was not associated with any recorded character string in step S904, the feature quantity extraction unit 192 converts the attribute of the character string (character string attribute) into a feature quantity. As in the case of the first feature amount, at the time of step S908 when the document forms are not grouped by format, it is not possible to generate a common feature amount for the document forms of the same format. Therefore, the feature amount extraction unit 192 extracts the feature amount for each document form and for each non-recorded character string in preparation for extracting the second feature amount for each group. The feature amount for each document form and for each non-recorded character string is called an individual second feature amount.
The feature amount extraction unit 192 may generate an individual second feature amount for each document form and by summarizing a plurality of non-recorded character strings. For example, the feature quantity extraction unit 192 may generate one individual second feature quantity for one document form.

The feature amount extraction unit 192 records the obtained individual second feature amount in the database 4 in association with the identifier of the document form and the identifier of the unrecorded character string (step S909). As the identifier of the non-recorded character string, for example, a coordinate value indicating the position of the non-recorded character string can be used.
For example, the feature quantity extraction unit 192 extracts individual characters indicating the name 501 of the orderer, the emblem image of the orderer, the title 503 of the document form, the greeting 504, etc., which are non-recorded character strings included in the format of the document form 5 in FIG. The second feature amount is recorded in the database 4 in association with the identifier of the document form 5 and the identifier of the unrecorded character string.

In the database 4, for each of a plurality of formats of the document form, image data of the document form of the format and information of recorded character strings corresponding to the image data are recorded. The acquisition unit 191 of the image processing apparatus 1 repeats the processing of steps S901 to S909 until the image data and recorded character string information for all document forms are read.

If it is determined in step S901 that the acquisition unit 191 has read all the image data of the document form and the information of the recorded character string corresponding to the image data from the database 4 (step S901: YES), the group classification unit 194 The forms are grouped (step S921). The grouping unit 194 groups the document forms based on the individual second feature amount included in the image data of the document forms. For example, the grouping unit 194 groups each document form based on the degree of matching of non-recorded character strings indicated by the individual second feature amount, the degree of matching of emblem images, the degree of matching of coordinate ranges of non-recorded character strings, and the like. do. The grouping unit 194 determines group identifiers of document forms in this grouping process. The grouping unit 194 determines whether all document forms have been grouped (step S922).

The grouping unit 194 repeats the process of step S921 if grouping of all the document forms has not been completed. Specifically, when the grouping unit 194 determines in step S922 that there is a document form that has not been grouped (step S922: NO), the process returns to step S921.
When grouping of all the document forms is completed (step S922: YES), the grouping unit 194 associates the identifier of the document form with the group identifier assigned to the document form, table) (step S923).

Then, the feature amount extraction unit 192 reads each individual first feature amount and each individual second feature amount of one or a plurality of document forms belonging to a certain group from the database 4, and extracts each individual first feature amount of the document forms belonging to the group. Each group first feature amount and each group second feature amount corresponding to the amount and each individual second feature amount are extracted (step S924). Each group first feature amount may be a value such as an average of individual first feature amounts of documents belonging to the group. Similarly, each group second feature amount may be a value such as the average of each individual second feature amount of the document form belonging to the group. Each group's first feature amount and each group's second feature amount are not the average of each individual first feature amount and the average of each individual second feature amount, but by using a method such as predetermined statistical processing or machine learning If it is a feature amount calculated so as to identify recorded character strings and non-recorded character strings of one or more document forms belonging to a group, each group's first feature amount, each group's first feature amount, each group You may extract two feature-values.

For example, in step S924, the feature amount extraction unit 192 may extract the character string attribute for each recorded character string from a plurality of document forms belonging to the same group and (directly) generate the first group feature amount. good. In this case, the feature quantity extraction unit 192 passes the extraction and recording of the individual first feature quantity in steps S905 and S906 (no particular processing is performed).
On the other hand, the second individual feature amount is extracted by the feature amount extraction unit 192 in step S908 so that it can be used for grouping in step S921. However, in step 921, the grouping unit 194 may group the document forms using non-recorded character strings instead of using the second individual feature amount. In this case, in step S924, the feature amount extraction unit 192 may (directly) extract the group second feature amount from the character string attributes of the non-recorded character strings of the plurality of document forms belonging to the same group. In this case, the feature quantity extraction unit 192 does not perform any particular processing in steps S907 to S909.
The feature amount extraction unit 192 calculates the first group feature amount and the second group feature amount for each group, and records them in the database 4 in association with the group identifier (step S925).
After step S925, the image processing apparatus 1 terminates the processing of FIG.

Through the above processing, the image processing apparatus 1 extracts the information (group first feature amount and group second feature amount) necessary for reducing the labor of the operator to record the record character string for each group of document forms. can be stored in the database 4. As a result, the image processing apparatus 1 can receive input of image data of a new document form and automatically record the record character string included in the new document form in the database 4 . The processing will be described with reference to FIG.

FIG. 12 is a second diagram showing the processing flow of the image processing apparatus according to the second embodiment. FIG. 12 shows an example of a processing procedure in which the image processing apparatus 1 extracts a recorded character string from newly input image data.
The operator performs an operation to cause the image reading device 2 to read a new document form. As a result, the image reading device 2 generates image data of the document form and outputs (transmits) it to the image processing device 1 . The acquisition unit 191 of the image processing apparatus 1 acquires image data from the data received by the communication unit 110 (step S1001). Acquisition unit 191 outputs the image data to feature amount extraction unit 192 . The feature amount extraction unit 192 performs OCR processing on the image data, and extracts a character string, characteristics of characters (character attributes) included in the character string, and coordinates in the image data within the range of the character string for each character string. Detect (step S1002). The feature amount extraction unit 192 extracts the third feature amount obtained by converting the detected information into the feature amount for each character string in the image data (step S1003). The third feature amount is information indicating the feature of the character string included in the newly read document form of the image data.

Next, the group specifying unit 195 reads out the group second feature amount to be used for group specification of the new document form among the group second feature amounts stored in the database 4 . The group second feature amount may be, for example, a feature amount corresponding to the emblem image 502 of the orderer displayed in the image data of the document form. The group identification unit 195 determines whether or not information indicated by a certain group second feature amount can be identified from the image data of the document form acquired in step S1001. The group identification unit 195 performs similar processing using the group second feature amount for all groups. If information matching the group second feature value read from the database 4 can be specified from the image data of the newly read document form, the group specifying unit 195 newly reads the group having the group second feature value. A group of image data of a document form is specified (step S1004). After that, the group identification unit 195 reads out one or more group first feature amounts for the group from the database 4 (step S1005). The group identification unit 195 outputs the third feature amount and one or more group first feature amounts to the recording unit 193 . The group first feature amount is a feature amount for specifying one or more recorded character strings in the document form belonging to the group.

The recording unit 193 acquires the third feature amount for each character string in the image data and the group first feature amount for each recorded character string, and associates the group first feature amount and the third feature amount (step S1006). . Specifically, the recording unit 193 associates each of the group first feature amounts with a third feature amount that matches the feature amount or one of the closest third feature amounts. With this association, a recorded character string is selected from the character strings obtained by OCR processing the image data of the document form.
Next, the feature quantity extraction unit 192 controls the display unit 120 to display the recorded character string, which is the result of processing by the image processing apparatus 1, and the image of the document form side by side (step S1007). Step S1007 is the same as step S706 in FIG.

Next, the recording unit 193 determines whether the correspondence relationship between the display of the recorded character string and the image of the document form has been obtained (step S1008). For example, the recording unit 193 determines that the correspondence has been acquired when the number of times of learning for grasping the correspondence between the display of the recorded character string and the image of the document form is equal to or greater than a predetermined number. As described above, the learning for grasping the correspondence relationship between the display of the recorded character string and the image of the document form may be the learning for specifying the recorded character string in the display of the recorded character string, or the learning for identifying the recorded character string in the display of the recorded character string. It may be learning for identifying the recorded character string in the image of the form. Alternatively, it may be a combination of these learnings.

If it is determined that the correspondence relationship between the display of the recorded character string and the image of the document form has been acquired (step S1008: YES), the recording unit 193 controls the display unit 120 to display the recorded character string and the image of the document form. The correspondence with the image is displayed (step S1009). Step S1009 is the same as step S708 in FIG.
When the operator confirms and corrects the recorded character string and then performs a confirmation operation, the image processing apparatus 1 acquires the corrected recorded character string (step S1010). If the recording unit 193 determines in step S1008 that the correspondence relationship between the display of the recorded character string and the image of the document form has not been obtained (step S1008: NO), the process also proceeds to step S1010.

Then, the feature quantity extraction unit 192 and the recording unit 193 as the correspondence relationship learning unit learn the coordinates of the recorded character string in the display of the processing result (in the example of FIGS. 8 and 9, the recorded character string display window W2). step S1011). For example, the feature amount extraction unit 192 extracts the first feature amount for display of the processing result in the same manner as in the document form, and the recording unit 193 uses the first feature amount to determine the position of the recorded character string. Identify and remember.
If the recording unit 193 already knows the display format of the processing result, the processing of step S1011 is unnecessary.

The recording unit 193 records the recording character string in the recording table in association with the identification information of the document form (step S1012).
For example, assume that a third feature amount a3, a third feature amount b3, a third feature amount c3, and a third feature amount d3 have been acquired from the image data of the document form. Then, the third feature amount a3 is the first feature amount a1 recorded in advance in the database, the third feature amount b3 is the first feature amount b1, the third feature amount c3 is the first feature amount c1, and the third feature amount It is assumed that the feature amount of the quantity d3 matches the first feature amount d1. In this case, the recording unit 193 records the character strings corresponding to the third feature amount a3, the third feature amount b3, the third feature amount c3, and the third feature amount d3 in the record table of the document form as recorded character strings. do. The character string corresponding to the third feature amount referred to here is the character string from which the third feature amount is extracted. When the operator corrects the recorded character string, the recording unit 193 records the corrected recorded character string in the recording table of the document form.
After step S1012, the image processing apparatus 1 ends the processing of FIG.

The image processing apparatus 1 may update the first feature amount in the document form by the process of FIG. For example, in step S1011, the image processing apparatus 1 may learn the first feature amount in the document form in addition to or instead of the machine learning of the first feature amount in the processing result display. In that case, the image processing apparatus 1 may perform the processing flow of FIG. 6 again. Alternatively, the image processing apparatus 1 may perform additional learning in step S1011 to update the first feature amount without reprocessing the data processed in FIG.
The image processing device 1 updates the first feature quantity by the processing of FIG. 12, thereby increasing the number of sample data and improving the accuracy of the first feature quantity, thereby increasing the accuracy with which the image processing device 1 extracts the recorded character string. is expected to improve. 12, the image processing apparatus 1 can also extract the newly added recorded character string from the image data, saving the operator the trouble of inputting the character string. It is expected that it will be saved.

According to the process shown in FIG. 12, the image processing apparatus 1 selects the type of a newly input document form based on the document form image data and recorded character strings for each of a plurality of formats recorded in advance by the operator. It is possible to automatically record the record character string in the image data of the document form without using it. Therefore, the image processing apparatus 1 can reduce the labor of the operator for recording the record character string in the document form. Also, the operator can refer to the display on the display unit 120 to check and correct the recorded character string that is the processing result of the image processing apparatus 1 . At this time, the display unit 120 indicates the correspondence between the character strings in the display of the processing result and the character strings in the image of the document form, so that the operator can easily understand the correspondence.

<Third embodiment>
As another example of the processing of the image processing apparatus 1, the operator may register groups of document forms in the image processing apparatus 1 in advance. For example, when registering image data of a document form in the past, the operator entered a group identifier according to the type of the document form and registered it in the database 4 in association with the image data of the document form. As a result, different kinds of forms are not mixed in the same group due to a processing error of the image processing apparatus 1 or the like, and the first feature amount can be extracted with high accuracy. In this case, the operator inputs the group of the document form at the time of registration, but for the new form, the group is specified using the second feature amount, as in step S1004.

<Fourth embodiment>
As another example of processing of the image processing apparatus 1, the image processing apparatus 1 not only groups document forms using the second feature amount, but also uses the first feature amount to The document form may be grouped by using the first feature amount together with . The first feature amount is the feature amount of the recorded character string. If the document form is of the same type, it is considered that the coordinates and character attributes of the recorded character string are the same. Grouping is possible. The first grouping is performed by the operator as shown in the fourth embodiment, and the new document form is grouped using the first feature amount in the processing of step S1004, thereby accurately recording in the OCR process. Strings can be read.
In this case, the acquisition unit 191 acquires a plurality of form image data and a recorded character string to be recorded among the character strings included in the form image data. Then, the grouping unit 194 groups the form image data based on the first feature amount. Then, the feature quantity extraction unit 192 extracts the recorded character string using the first feature quantity corresponding to the form image data included in the group.

<Fifth embodiment>
In the second embodiment, in step S1004, a new group of forms is specified based on the second feature amount. However, as another processing mode, the image processing apparatus 1 does not perform the process of specifying groups, but sequentially specifies each group for all groups set by the operator, reads out the first feature amount, Count the number of matches with the third feature amount. In the case of the correct group, the first feature amount and the third feature amount should match the most. Record. This makes it possible to record the recording string without specifying the group.
In this case, the acquisition unit 191 acquires a plurality of form image data and a recorded character string to be recorded among the character strings included in the form image data. Then, the feature amount extraction unit 192 performs the first feature amount indicating the feature of the recorded character string or the second feature amount indicating recognition information other than the recorded character string based on the result of the recognition processing of the form image data acquired by the acquisition unit 191 . Extract features. The feature amount extraction unit 192 extracts a recorded character string using a first feature amount corresponding to form image data included in a predetermined group.

As described above, the recording unit 193 preliminarily records based on the results of learning using a plurality of document images. , a character string of a specific item in the first document image is detected based on the feature amount of the displayed first document image. The display unit 120 outputs correspondence information indicating the same specific item between the first document image and the second document image displayed corresponding to the first document image.
According to the image processing apparatus 1, the position of the character string in the document image can be acquired by learning. Therefore, according to the image processing apparatus 1, even if the format of the document to be subjected to character recognition is not known in advance, it is possible to assist the operator in understanding which item is shown where.

Also, the recording unit 193 detects a character string of the same item as the specific item in the first document image from the second document image.
Thereby, the image processing apparatus 1 can present the correspondence relationship between the character string of the specific item in the first document image and the character string of the specific item in the second document image to the user. The user can relatively easily check whether the character string is correct or incorrect by referring to the presentation of the correspondence.

Also, the feature amount extraction unit 192 and the recording unit 193 machine-learn the correspondence relationship of the same specific item between the first document image and the second document image.
As a result, the image processing apparatus 1 can automatically learn the correspondence relationship of the same specific item between the first document image and the second document image without requiring any special action by the user. The user only has to perform normal processing, and the burden on the user does not increase.

In addition, the feature amount extracting unit 192 and the recording unit 193 extract the same specific item in the first document image and the second document image using the image after the character string of the second document image is determined. machine-learning the correspondence between
If the character string in the second document image is confirmed, the confirmed character string is considered correct. By performing machine learning using this accurate character string, the correspondence learning unit can learn the correspondence of the same specific item between the first document image and the second document image with relatively high accuracy. is expected.

Next, the configuration of the embodiment of the present invention will be described with reference to FIG.
FIG. 12 is a diagram illustrating an example of the configuration of an image processing apparatus according to the embodiment; An image processing device 600 shown in FIG. 12 includes a character string detection unit 601 and an output unit 602 .
With such a configuration, the character string detection unit 601 obtains a feature amount that is recorded in advance based on the results of learning using a plurality of document images, and indicates the character string characteristics of the item for each type of document image and for each specific item. Among them, the character string of the specific item in the first document image is detected based on the feature amount of the displayed first document image.
The output unit 602 outputs correspondence information indicating the same specific item between the first document image and the second document image displayed corresponding to the first document image.
According to the image processing device 600, the position of the character string in the document image can be acquired by learning. Therefore, according to the image processing apparatus 600, even if the format of the document to be subjected to character recognition is not known in advance, it is possible to assist the operator in understanding which item is shown where.

Each of the devices described above has an internal computer system. A program for causing each device to perform each process described above is stored in a computer-readable recording medium of each device, and the computer of each device reads and executes the program to perform the above process done. Here, the computer-readable recording medium refers to magnetic disks, magneto-optical disks, CD-ROMs, DVD-ROMs, semiconductor memories, and the like. Alternatively, the computer program may be distributed to a computer via a communication line, and the computer receiving the distribution may execute the program.

Further, the program may be for implementing part of the functions of the processing units described above. Further, it may be a so-called difference file (difference program) that can realize the above-described functions in combination with a program already recorded in the computer system.

Although the embodiment of the present invention has been described in detail with reference to the drawings, the specific configuration is not limited to this embodiment, and design and the like are included within the scope of the gist of the present invention.

1 image processing device 2 image reading device 3 recording device 4 database 110 communication unit 120 display unit 130 operation input unit 180 storage unit 190 control unit 191 acquisition unit 192 feature amount extraction unit 193 recording unit

Claims (8)

a character string recognition unit that recognizes the character string of a specific item in an image to be recognized based on the results of learning using a plurality of images containing character strings;
Output for outputting the recognition target image and the character string recognition result in a manner in which a correspondence relationship indicating the same specific item in the recognition target image and the character string recognition result by the character string recognition unit can be grasped. Department and
with
The character string recognition unit identifies candidates for the character string of the specific item in the image to be recognized,
The output unit outputs the range of the identified character string candidates in a comprehensible manner.
Image processing device. an operation reception unit that receives an operation to correct a character string included in the character string recognition result output by the output unit;
The image processing apparatus according to claim 1, wherein the character string recognition unit records the character string corrected by the correction operation accepted by the operation accepting unit as the recognized character string. The character string recognizing unit selects the character string for each type of document indicated by the plurality of images including the character string and for each of the specific items from among the feature amounts that are recorded based on the learning result and indicate the characteristics of the character string for each of the specific items. 3. The image processing apparatus according to claim 1, wherein the character string of the specific item is recognized based on the feature amount of the character string in the target image. The image processing apparatus according to claim 3, wherein the feature amount is generated based on information indicating attributes of characters included in the character string and a position of the character string in the image to be recognized. 5. The image processing device according to claim 4, wherein the character attribute indicates at least one of numbers, alphabets, hiragana, kanji, number of characters, character height, and font. The character string recognition unit detects a plurality of candidates for the character string of the specific item in the recognition target image,
The image processing apparatus according to any one of claims 1 to 5, wherein the output unit outputs a range of each of the detected character string candidates in a manner that allows understanding. a step of recognizing the character string of a specific item in an image to be recognized based on the results of learning using a plurality of images containing the character string;
a step of outputting the recognition target image and the character string recognition result in a manner in which a correspondence relationship indicating the same specific item in the recognition target image and the character string recognition result can be grasped;
including
In the step of recognizing the character string, identifying candidates for the character string of the specific item in the image to be recognized;
In the outputting step, the range of the identified character string candidates is output in a comprehensible manner.
Image processing method. to the computer,
a step of recognizing the character string of a specific item in an image to be recognized based on the results of learning using a plurality of images containing the character string;
a step of outputting the recognition target image and the character string recognition result in a manner in which a correspondence relationship indicating the same specific item in the recognition target image and the character string recognition result can be grasped;
and
In the step of recognizing the character string, identifying candidates for the character string of the specific item in the image to be recognized;
In the outputting step, the range of the identified character string candidates is output in a comprehensible manner.
program for.

Images