Image processing system, image processing method and image processing program

Abstract

To provide an image processing system, an image processing method and an image processing program capable of reducing a throughput of recognition processing to a prescribed item of an input image and improving recognition accuracy.SOLUTION: An acquisition processing part 111 acquires a result of character recognition processing which is executed to an object character string of at least any of a name and a corporation name in an image. An evaluation processing part 114 calculates evaluation values for multiple registered character strings, on the basis of set values set to multiple character string candidates included in a result of the character recognition processing, and registered values each of which is related to each registered character string corresponding to the object character string and registered in advance. A determination processing part 115 identifies a prescribed registered character string on the basis of the respective evaluation values of the multiple registered character strings. An output processing part 116 outputs the prescribed registered character string.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a technique for performing image processing for character recognition on images such as documents. [Background technology]

[0002] Conventionally, there are known techniques for detecting character strings from images of documents such as documents, slips, etc. For example, when detecting a destination from an input image, a technique is known in which the input image is divided into predetermined unit block areas, features are generated for each unit block area, unit block areas other than the destination are excluded, and the destination is detected from the remaining unit block areas (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 1-321587 Summary of the Invention [Problem to be solved by the invention]

[0004] However, in the case of input images with a large amount of information, such as documents, the number of blocks to be divided increases, resulting in a large amount of processing. In addition, it becomes difficult to generate features for each block, resulting in a decrease in recognition accuracy.

[0005] An object of the present disclosure is to provide an image processing system, an image processing method, and an image processing program that are capable of reducing the amount of processing for recognition of specified items of an input image and improving recognition accuracy. [Means for solving the problem]

[0006] An image processing system according to one aspect of the present disclosure includes an acquisition processing unit, an evaluation processing unit, a specification processing unit, and an output processing unit. The acquisition processing unit acquires a result of a character recognition process executed on at least one target character string of a name and / or a corporate name included in an image. The evaluation processing unit calculates an evaluation value for each of a plurality of registered character strings based on a setting value set for each of a plurality of character string candidates included in the result of the character recognition process and a registration value associated with each of a plurality of registered character strings that are previously registered and correspond to the target character string. The specification processing unit identifies a predetermined registered character string based on the evaluation value for each of the plurality of registered character strings calculated by the evaluation processing unit. The output processing unit outputs the predetermined registered character string identified by the specification processing unit.

[0007] An image processing method according to another aspect of the present disclosure is a method executed by one or more processors to obtain the results of a character recognition process performed on at least one target character string of a name and / or corporate name included in an image, calculate an evaluation value for each of a plurality of registered character strings based on a setting value set for each of a plurality of candidate character strings included in the result of the character recognition process and a registered value associated with each of a plurality of registered character strings that correspond to the target string, identify a specified registered character string based on the evaluation value for each of the plurality of registered character strings, and output the specified registered character string.

[0008] An image processing program according to another aspect of the present disclosure is a program for causing one or more processors to execute the following steps: obtain the result of a character recognition process performed on at least one target character string of a name and / or corporate name included in an image; calculate an evaluation value for each of a plurality of registered character strings based on a setting value set for each of a plurality of candidate character strings included in the result of the character recognition process and a registered value associated with each of a plurality of registered character strings that correspond to the target string and are registered in advance; identify a specific registered character string based on the evaluation value for each of the plurality of registered character strings; and output the specific registered character string. Effect of the Invention

[0009] According to the present disclosure, it is possible to provide an image processing system, an image processing method, and an image processing program that are capable of reducing the amount of processing for recognition of specified items of an input image and improving recognition accuracy. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a functional block diagram showing a configuration of an image processing device according to an embodiment of the present disclosure. [Diagram 2] FIG. 2 is a diagram illustrating an example of a document (quote) according to an embodiment of the present disclosure. [Diagram 3] FIG. 3 is a diagram illustrating an example of a rectangle detected from image data according to an embodiment of the present disclosure. [Figure 4] FIG. 4 is a diagram illustrating an example of a keyword table according to an embodiment of the present disclosure. [Diagram 5] FIG. 5 is a diagram illustrating an example of blocks obtained by dividing an image according to an embodiment of the present disclosure. [Figure 6] FIG. 6 is a diagram showing telephone number and address information detected from an image according to an embodiment of the present disclosure. [Figure 7] FIG. 7 is a diagram showing a state in which telephone numbers and address locations are plotted on blocks according to an embodiment of the present disclosure. [Figure 8] FIG. 8 is a flowchart showing an example of a predetermined range set in an image according to an embodiment of the present disclosure. [Figure 9] FIG. 9 is a diagram illustrating an example of calculation of the evaluation value according to the embodiment of the present disclosure. [Figure 10] FIG. 10 is a diagram illustrating an example of calculation of the evaluation value according to the embodiment of the present disclosure. [Figure 11] FIG. 11 is a diagram illustrating an example of calculation of the evaluation value according to the embodiment of the present disclosure. [Figure 12A] FIG. 12A is a diagram illustrating an example of calculation of an evaluation value corresponding to the keyword "supplier" according to an embodiment of the present disclosure. [Figure 12B] FIG. 12B is a diagram illustrating an example of calculation of an evaluation value corresponding to the keyword "supplier" according to an embodiment of the present disclosure. [Figure 13A] FIG. 13A is a diagram showing an example of calculation of an evaluation value corresponding to a keyword "company name" according to an embodiment of the present disclosure. [Figure 13B] FIG. 13B is a diagram showing an example of calculation of an evaluation value corresponding to a keyword "company name" according to an embodiment of the present disclosure. [Figure 14A] FIG. 14A is a diagram showing an example of calculation of an evaluation value corresponding to a keyword "department name" according to an embodiment of the present disclosure. [Figure 14B] FIG. 14B is a diagram showing an example of calculation of an evaluation value corresponding to a keyword "department name" according to an embodiment of the present disclosure. [Figure 15A] FIG. 15A is a diagram illustrating a calculation example of an evaluation value corresponding to a keyword "title" according to an embodiment of the present disclosure. [Figure 15B] FIG. 15B is a diagram illustrating an example of calculation of an evaluation value corresponding to a keyword "title" according to an embodiment of the present disclosure. [Figure 16] FIG. 16 is a diagram showing an example of calculation of an evaluation value corresponding to the keyword "transfer destination" according to an embodiment of the present disclosure. [Figure 17] FIG. 17 is a diagram illustrating an example of calculation of an evaluation value corresponding to the keyword "account" according to an embodiment of the present disclosure. [Figure 18] FIG. 18 is a diagram illustrating an example of calculation of evaluation values ​​corresponding to the types of characters in a collocation according to an embodiment of the present disclosure. [Figure 19] FIG. 19 is a diagram illustrating an example of calculation of an evaluation value corresponding to a type of text according to an embodiment of the present disclosure. [Figure 20] FIG. 20 is a diagram illustrating an example of error correction values ​​for OCR candidates according to an embodiment of the present disclosure. [Figure 21] FIG. 21 is a diagram illustrating an example of calculation of an evaluation value corresponding to an address feature according to an embodiment of the present disclosure. [Figure 22] FIG. 22 is a diagram illustrating an example of calculation of an evaluation value corresponding to an address feature according to an embodiment of the present disclosure. [Figure 23] FIG. 23 is a diagram illustrating an example of calculation of an evaluation value corresponding to a registration number according to an embodiment of the present disclosure. [Figure 24] FIG. 24 is a flowchart showing an example of a procedure of image processing executed in the image processing device according to an embodiment of the present disclosure. [Diagram 25] FIG. 25 is a diagram showing an example of an OCR result used in the surname evaluation process according to an embodiment of the present disclosure. [Figure 26] FIG. 26 is a diagram showing an example of a surname table used in the surname evaluation process according to an embodiment of the present disclosure. [Figure 27] FIG. 27 is a diagram showing an example of calculation of an evaluation value in the surname evaluation process according to an embodiment of the present disclosure. [Figure 28] FIG. 28 is a diagram showing an example of a name table used in the name evaluation process according to an embodiment of the present disclosure. [Figure 29A] FIG. 29A is a diagram showing an example of an OCR result used in the name evaluation process according to an embodiment of the present disclosure. [Figure 29B] FIG. 29B is a diagram showing an example of calculation of an evaluation value in the name evaluation process according to an embodiment of the present disclosure. [Figure 29C] FIG. 29C is a diagram showing an example of calculation of an evaluation value in the name evaluation process according to an embodiment of the present disclosure. [Diagram 30] FIG. 30 is a diagram illustrating an example of a word table used in the corporate name evaluation process according to an embodiment of the present disclosure. [Diagram 31] FIG. 31 is a diagram illustrating an example of a character table used in the corporate name evaluation process according to an embodiment of the present disclosure. [Figure 32A] FIG. 32A is a diagram showing an example of calculation of an evaluation value in the corporate name evaluation process according to an embodiment of the present disclosure. [Figure 32B] FIG. 32B is a diagram showing an example of calculation of an evaluation value in the corporate name evaluation process according to an embodiment of the present disclosure. [Figure 32C] FIG. 32C is a diagram showing an example of calculation of an evaluation value in the corporate name evaluation process according to an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. Note that the following embodiments are examples of the present disclosure and are not intended to limit the technical scope of the present disclosure.

[0012] [Image processing device 1] 1 is a block diagram showing a configuration of an image processing device 1 according to an embodiment of the present disclosure. The image processing device 1 recognizes a character string included in an input image (image data), performs a well-known character recognition process (OCR process) on the recognized character string, and further performs a predetermined image process on the result of the OCR process, and outputs the process result. Note that the image processing device 1 may obtain the result of the OCR process performed in an external device and perform the predetermined image process.

[0013] 1, the image processing device 1 includes a control unit 11, a storage unit 12, an operation display unit 13, and a communication unit 14. The image processing device 1 is an information processing device such as a personal computer or a cloud server. The image processing device 1 may be one or more cloud servers, or one or more physical servers.

[0014] The communication unit 14 is a communication interface for connecting the image processing device 1 to a network in a wired or wireless manner and for executing data communication with an external device via the network in accordance with a predetermined communication protocol. The network may be, for example, the Internet or a LAN.

[0015] The operation display unit 13 is a user interface that includes a display unit, such as a liquid crystal display or an organic EL display, that displays various information, and an operation unit, such as a mouse, a keyboard, or a touch panel, that accepts operations.

[0016] The storage unit 12 is a non-volatile storage unit such as an HDD (Hard Disk Drive), an SSD (Solid State Drive), or a flash memory that stores various information. The storage unit 12 stores a control program such as an image processing program (one example of an image processing program of the present disclosure) for causing the control unit 11 to execute image processing (see FIG. 24) described below. For example, the image processing program is non-temporarily recorded on a computer-readable recording medium such as a CD or a DVD, and is read by a reading device (not shown) such as a CD drive or a DVD drive provided in the image processing device 1 and stored in the storage unit 12. The image processing program may be distributed from a cloud server and stored in the storage unit 12.

[0017] The storage unit 12 also stores image data (scanned data, etc.) of documents and the like acquired from external devices.

[0018] FIG. 2 shows an estimate as an example of a document. As shown in FIG. 2, an estimate includes multiple character strings such as document classification ("estimate"), issue date, destination, issuer (company name, postal code, address, telephone number, registration number, etc.), estimated amount, product name, quantity, unit price, amount, and transfer account. A user scans the estimate using an external device and uploads image data (input image) to the image processing device 1. When the control unit 11 acquires image data of the estimate, it stores the image data in the storage unit 12. As another embodiment, the control unit 11 may acquire a document file of the estimate created in an external device and store the document file in the storage unit 12.

[0019] The control unit 11 has control devices such as a CPU, a ROM, and a RAM. The CPU is a processor that executes various arithmetic operations. The ROM stores in advance control programs such as a BIOS and an OS for causing the CPU to execute various processes. The RAM stores various information and is used as a temporary storage memory (work area) for the various processes executed by the CPU. The control unit 11 controls the image processing device 1 by executing, by the CPU, various control programs that are stored in advance in the ROM or the storage unit 12.

[0020] When the control unit 11 acquires image data of a document, it performs preprocessing such as resolution conversion processing, skew correction processing, and background removal processing on the image data, and then performs character string (document part) recognition processing (extraction processing). For example, the control unit 11 recognizes each part such as character strings, tables, illustrations, and seals as rectangular areas of objects in the image data P1 (see FIG. 3) on which the preprocessing has been performed. Furthermore, after performing document part recognition processing, the control unit 11 performs character recognition processing (OCR processing) on ​​the entire image data (entire image). Furthermore, the control unit 11 performs a predetermined process on the result of the OCR processing.

[0021] In this embodiment, the control unit 11 executes OCR processing and determination processing on character strings of predetermined items included in the input image. The predetermined items include at least one of the issuer, destination, and registration number. In the following, "issuer" will be described as an example of the predetermined item.

[0022] Specifically, as shown in Fig. 1, the control unit 11 includes various processing units such as an acquisition processing unit 111, a collocation processing unit 112, a detection processing unit 113, an evaluation processing unit 114, a judgment processing unit 115, and an output processing unit 116. The control unit 11 functions as the various processing units by executing various processes according to the image processing program. Some or all of the processing units included in the control unit 11 may be configured with electronic circuits. The image processing program may be a program for causing a plurality of processors to function as the various processing units.

[0023] The acquisition processing unit 111 acquires an input image. Specifically, the acquisition processing unit 111 acquires an image (input data) to be subjected to character recognition from an external device. For example, the acquisition processing unit 111 acquires image data P1 (see FIG. 3) of various document images such as an estimate shown in FIG. 2. The acquisition processing unit 111 also executes OCR processing on the acquired image data P1 to generate character string rectangles, font size, and character information. As another embodiment, the acquisition processing unit 111 may acquire searchable PDF data. When the acquisition processing unit 111 acquires searchable PDF data, the control unit 11 transmits the image data to the OCR processing unit, and transmits the words, rectangle positions, font size, and character information of the searchable PDF to the collocation processing unit.

[0024] The collocation processor 112 executes collocation processing to create character strings (compound words) for each item from the image data P1. For example, the collocation processor 112 detects character string rectangles in the image data P1, and performs processing to determine and remove garbage data, processing to divide data of two or more lines into one line unit, processing to connect based on the distance between characters, and the like to create collocations. For example, since the OCR result contains a mixture of characters, words, and compound words, the collocation processor 112 appropriately collates these. For example, the collocation processor 112 compares font sizes in the horizontal direction and connects rectangles with close distances between them to create collocations.

[0025] The detection processing unit 113 detects a specific character string of a related item related to the detection target among the character strings that are character-recognized and are included in the image data P1. Specifically, the detection processing unit 113 executes a collocation process, and then executes an extraction process to extract information of each item. For example, the detection processing unit 113 executes processes to extract keywords, NG words, registration numbers, postal codes, telephone numbers, corporate names, addresses, national agencies, departments, reserve candidates, marks, form types, and the like in order. The detection target is, for example, an issuer or destination included in a form image, and the specific character string is at least one of a postal code, an address, and a telephone number.

[0026] For example, the detection processing unit 113 detects a keyword (specific character string) that is important for detecting a detection target such as an issuer, a destination, a registration number, etc. The detection processing unit 113 detects a keyword by referring to a keyword table shown in FIG.

[0027] Furthermore, the detection processing unit 113 detects and deletes a collocation that includes a clearly inapplicable word (NG word), such as "publication date." The detection processing unit 113 detects an NG word, for example, by referring to an NG word table.

[0028] In addition, the detection processing unit 113 detects a registration number based on the invoice system. Specifically, the detection processing unit 113 detects "T+13 digits".

[0029] Furthermore, the detection processing unit 113 detects a postal code. Specifically, the detection processing unit 113 detects a postal code based on whether or not a keyword such as "〒" or "postal code" matches a seven-digit number with a postal code. For example, when a collocation is formed such as "492 Minosho-cho, Yamatokoriyama City, Nara Prefecture, 639-1186," the detection processing unit 113 detects "639-1186" as a postal code, deletes "〒639-1186" from the collocation, and leaves "492 Minosho-cho, Yamatokoriyama City, Nara Prefecture" as the collocation.

[0030] Furthermore, the detection processing unit 113 detects a telephone number based on whether or not a keyword such as "telephone" or "FAX" matches the area code and local code.

[0031] Furthermore, when a character string such as "Kaisho Kaisha" or "Yugen Kaisha" is included in a collocation, the detection processing unit 113 detects it as a corporate name.

[0032] Furthermore, the detection processing unit 113 detects the address by referring to the address table and checking whether or not there is a match.

[0033] Furthermore, the detection processing unit 113 detects the name of a national institution by using a keyword for the name of the national institution, and detects the name of a department by using a keyword for the name of the department.

[0034] In addition, the detection processing unit 113 detects, as reserve candidates, those excluding those that cannot be a destination or issuer, such as NG words. For example, the detection processing unit 113 detects, as reserve candidates, those collocations that have not been detected as an address, a corporation, etc., that may be a source or destination, among the detected collocations. The reserve candidates include names, corporation names, etc.

[0035] Furthermore, the detection processing unit 113 detects marks that may be included in a company name or the like as a key mark and cannot be acquired in advance.

[0036] Furthermore, for forms such as "quotation," "invoice," "delivery note," "order form," and "receipt," the detection processing unit 113 detects and classifies the title in the upper center section. If there is no title in the upper center section, the detection processing unit 113 classifies it as "other."

[0037] As described above, the detection processing unit 113 detects from the collocation data what can be determined as a keyword, a postal code, an address, a telephone number, etc. At that time, the detection processing unit 113 also executes processes such as cutting the collocation and connecting adjacent characters.

[0038] The evaluation processing unit 114 calculates the evaluation value of the other character strings based on the position of the specific character string (such as a postal code, address, or phone number) in the image data P1 and the positions of the other character strings excluding the specific character string. Specifically, the evaluation processing unit 114 divides the area of ​​the image data P1 into a plurality of blocks, and calculates the evaluation value based on the positional relationship between a first block corresponding to the position of the specific character string (such as a postal code, address, or phone number) and a second block corresponding to the position of the other character strings. The evaluation processing unit 114 also calculates the evaluation value based on the total number of the specific character strings included in a predetermined range set according to the detection target (for example, issuer, destination, or registration number) with the position of the first block as the center.

[0039] Specifically, the evaluation processing unit 114 calculates the evaluation values ​​of the likelihood of the issuer, destination, and registration number from the position of the collocation and the correlation of the judgment contents. In the following, as an example, a case where the evaluation processing unit 114 estimates the "issuer" will be described.

[0040] The evaluation processing unit 114 divides the area of ​​the form image into a plurality of blocks. For example, the evaluation processing unit 114 calculates the aspect ratio of the area of ​​the form where characters are present, and calculates the number of divisions in the vertical and horizontal directions according to the aspect ratio. For example, when the aspect ratio is between 1:2 and 2:1, the evaluation processing unit 114 divides the area of ​​the form where characters are present into 16×16 blocks B1 (see FIG. 5). The evaluation processing unit 114 expresses the block B1 in which each collocation is located by using coordinates. For example, when the 16×16 block B1 is expressed by the coordinates (0,0) to (15,15), the evaluation processing unit 114 determines the coordinates of the block B1 in which the center position of each collocation is located. The evaluation processing unit 114 determines the position of each collocation in which block B1.

[0041] The evaluation processing unit 114 also determines the type of document. For example, if a keyword is found in the first three blocks from the top, the evaluation processing unit 114 determines that the keyword is the type of form. If multiple keywords exist, the evaluation processing unit 114 identifies multiple types (e.g., "estimate and invoice").

[0042] Next, the evaluation processing unit 114 calculates an evaluation value representing the likelihood of the source and the likelihood of the destination for each collocation, in order to select the source and the destination from among the collocations. Here, since postal codes, telephone numbers, and addresses are often located near the source and the destination, respectively, the evaluation processing unit 114 calculates a higher evaluation value the closer the target collocation is to the postal code, telephone number, or address. In addition, the evaluation processing unit 114 calculates an evaluation value for each block B1 regarding the positional relationship of the address, and sets the same evaluation value to the collocations in that block B1.

[0043] First, the evaluation processing unit 114 corrects the evaluation value of the collocation based on the positional relationship with the postal code, telephone number, and address. Specifically, the evaluation processing unit 114 compares the place names represented by the detected postal code, telephone number, and address. For example, as shown in Fig. 6, assume that there are three collocations (marks 1 to 3) that include the same place name ("Waki-cho, Okayama Prefecture").

[0044] The evaluation processing unit 114 plots marks 1 to 3 on a 16×16 block B1 as shown in FIG. 7. Next, as shown in FIG. 8, the evaluation processing unit 114 counts the number of collocations of the same place name contained in a 5×5 area A1 (predetermined range) centered on each block B1, and calculates the maximum value for each block B1. In the example of FIG. 8, the evaluation processing unit 114 calculates the maximum value "3" for block B1 at (10,2). The size of area A1 is set according to the total number (number of divisions) of blocks B1.

[0045] Next, for the sake of convenience, the evaluation processing unit 114 doubles the value of each block B1, and then compares the value of its own block B1 with half of the adjacent blocks B1 on the top, bottom, left, and right, and updates the value to the largest one. For example, the evaluation processing unit 114 sets the value of blocks B1 (11,1) and (10,2) that contain place name collocations to "6". Focusing on block B1 (11,2), the block B1 itself (11,2) is "0" because there is no place name collocation, the blocks B1 on the left (11,2) and above (10,1) are "3" which is half of "6", and the blocks B1 on the right (12,2) and below (11,3) are "0" because there is no place name collocation. Therefore, the evaluation processing unit 114 updates the evaluation value of block B1 (11,2) to "3" as shown in FIG. 9.

[0046] Next, the evaluation processing unit 114 corrects the evaluation value by the address. For example, since the characters of the issuer are often located outside the upper left area of ​​the document, the evaluation processing unit 114 multiplies the previously calculated evaluation value of the address by a predetermined coefficient p1 (for example, "3.5") for each block B1 other than the upper left area, as shown in FIG.

[0047] Next, the evaluation processing unit 114 calculates an evaluation value according to the position of the collocation, as shown in FIG. 11 (position information evaluation process).

[0048] For example, the evaluation processing unit 114 sets a high evaluation value to the upper right block B1 where the source character is highly likely to exist. The evaluation values ​​in FIG. 11 may be a fixed table that is set in advance.

[0049] Next, the evaluation processing unit 114 calculates an evaluation value corresponding to the keyword "Supplier" in "Delivery note." For example, as shown in Figures 12A and 12B, when the classification of a document is "Delivery note," the evaluation processing unit 114 sets an evaluation value of "44" to the collocation closest to the bottom right of "Supplier," and sets an evaluation value of "22" to the collocation next closest to the bottom right of "Supplier."

[0050] Next, the evaluation processing unit 114 calculates an evaluation value for the keyword "company name." For example, as shown in Figures 13A and 13B, when a corporate name, department name, or reserve candidate that includes the characters for "large company" is in the top left block B1, the evaluation processing unit 114 assigns an evaluation value of "-15" to the collocation in block B2.

[0051] Next, the evaluation processor 114 calculates an evaluation value corresponding to the keyword "department name." For example, as shown in Figures 14A and 14B, when a word containing the characters of the department name is found in the bottom right block B1, the evaluation processor 114 assigns an evaluation value of "10" to the collocation in block B2.

[0052] Next, the evaluation processing unit 114 calculates an evaluation value corresponding to the keyword "title". For example, as shown in Fig. 15A and Fig. 15B, when a title exists in block B2, the evaluation processing unit 114 assigns an evaluation value to each block B1 within one block above block B2 and within eight blocks to the left of block B2. For example, when the title type is "Onchu", the evaluation processing unit 114 assigns an evaluation value of "-20", and when the title type is "Dono, Sama", the evaluation processing unit 114 assigns an evaluation value of "-10".

[0053] Next, the evaluation processing unit 114 calculates an evaluation value corresponding to the keyword "transfer destination." For example, as shown in Fig. 16, when the keyword "transfer destination" exists in block B2, the evaluation processing unit 114 assigns the evaluation value shown in Fig. 16 to each block B1 in the shaded area.

[0054] Next, the evaluation processing unit 114 calculates an evaluation value corresponding to the keyword "account." For example, as shown in Fig. 17, when an account name, account type, etc. are present in block B1 of the shaded area, the evaluation processing unit 114 assigns an evaluation value of "-10" to the collocation in block B2. Note that since information such as an account number may be present after the account name or deposit type, the horizontal impact is large, but the range of impact in the vertical direction is small due to the layout in columns.

[0055] The evaluation processor 114 calculates the sum of the above evaluation values ​​for each collocation.

[0056] Next, the evaluation processing unit 114 calculates an evaluation value based on the characteristics of the collocation itself (collocation evaluation process). Specifically, the evaluation processing unit 114 calculates an evaluation value corresponding to the font size. For example, as shown in Fig. 18, when a collocation is a character, the evaluation processing unit 114 assigns an evaluation value of the font size to the collocation. Also, when a label is an illustration, the evaluation processing unit 114 assigns an evaluation value of the label that is twice as large in width and height.

[0057] Next, the evaluation processing unit 114 calculates an evaluation value corresponding to the type of text. For example, as shown in FIG. 19, the evaluation processing unit 114 assigns an evaluation value according to the type and item of national institution, corporation, department, and reserve candidate. In FIG. 19, the OCR error correction is a process in which multiple candidates are output for each character in the OCR output. When detecting a keyword, the value shown in FIG. 20 is added each time a second or lower candidate of the OCR is mixed in. When there is only one candidate, such as in a searchable PDF, processing is performed with only the first candidate. As another embodiment, the evaluation processing unit 114 may calculate an evaluation value for each of the name and the corporate name in the reserve candidate. The evaluation value of the name and the evaluation value of the corporate name are used as the determination elements of the issuer and the destination. The calculation method of the evaluation value of the name and the corporate name will be described later. Next, the evaluation processing unit 114 calculates an evaluation value corresponding to the features of the address. For example, if a collocation has 24 characters or more and the characteristic evaluation value of the address (see FIG. 22) is 4 or more, the evaluation processing unit 114 assigns an evaluation value of "-(address specific evaluation value-2) x 5", as shown in FIG. 21. FIG. 22 shows an example of a characteristic evaluation value table. The "address specific evaluation value" is the sum of the characteristic evaluation values ​​of the items included in the collocation.

[0058] Next, the evaluation processing unit 114 calculates an evaluation value corresponding to the registration number. For example, as shown in Fig. 23, when a predetermined condition is satisfied, the evaluation processing unit 114 assigns an evaluation value of "40" to the closest collocation. When a predetermined condition is not satisfied, the evaluation processing unit 114 assigns an evaluation value of "10" to the closest four or fewer collocations.

[0059] In the above example, the evaluation processing unit 114 calculates the evaluation value for the issuer, but it is also possible to calculate evaluation values ​​for the destination and registration number in the same way.

[0060] As described above, the evaluation processing unit 114 calculates an evaluation value indicating the likelihood of the issuer, destination, and registration number of the collocation itself. The evaluation processing unit 114 also calculates a higher evaluation value for the collocation, the closer it is to a specific character string (e.g., postal code, address, telephone number). The evaluation processing unit 114 also calculates the evaluation value based on the positional relationship between the first block corresponding to the position of the specific character string (postal code, address, telephone number, etc.) and the second block corresponding to the position of the other character string, the position of the first block in the image data P1, and the position of a specific character string (address, supplier, company name, department name, transfer destination, title, account, etc.). The evaluation processing unit 114 also calculates the evaluation value based on the characteristics of the collocation itself (font size, type of text, characteristics of the address, characteristics of the registration number, etc.).

[0061] The judgment processing unit 115 judges the ranking of the likelihood of each of the issuer, destination, and registration number by comparing the evaluation value of each collocation, which is the sum of the result of the location information evaluation process and the result of the collocation evaluation process.

[0062] The output processing unit 116 outputs, as a detection target character string, a character string identified based on the evaluation value calculated by the evaluation processing unit 114 from among the character strings (complex words). Specifically, the output processing unit 116 outputs the complex word having the highest evaluation value for each of the issuer, destination, and registration number.

[0063] [Image processing] FIG. 24 is a flowchart showing an example of the procedure of image processing executed in the image processing device 1 according to this embodiment.

[0064] The present disclosure can be understood as an image processing method (image processing method of the present disclosure) that executes one or more steps included in the image processing. One or more steps included in the image processing described here may be omitted as appropriate. The steps in the image processing may be executed in a different order as long as the same action and effect is achieved. Furthermore, although an example is described here in which the control unit 11 of the image processing device 1 executes each step in the image processing, in other embodiments, one or more processors may execute each step in the image processing in a distributed manner.

[0065] In step S1, the control unit 11 acquires an input image. For example, the control unit 11 acquires image data P1 of a document image (quote) shown in FIG.

[0066] Next, in step S2, the control unit 11 executes OCR processing. Specifically, the control unit 11 detects character string rectangles, which are areas of each character string, from the image data P1 (see FIG. 3), and outputs the character string for each character string rectangle.

[0067] Next, in step S3, the control unit 11 executes a collocation process. Specifically, the control unit 11 performs a process of determining and removing garbage data, a process of dividing data of two or more lines into one line unit, a process of connecting based on the distance between characters, and the like to create collocations.

[0068] Next, in step S4, the control unit 11 executes an information extraction process. Specifically, the control unit 11 executes processes to extract keywords, NG words, registration numbers, postal codes, telephone numbers, corporate names, addresses, national agencies, departments, reserve candidates, marks, document types, and the like from the collocated data.

[0069] Next, in step S51, the control unit 11 executes a position information evaluation process. Specifically, the control unit 11 calculates the aspect ratio of an area of ​​the document where characters are present, calculates the number of divisions in the vertical and horizontal directions based on the aspect ratio, and divides the document (image data P1) into a plurality of blocks B1 according to the number of divisions (see FIG. 5). The control unit 11 also determines in which block B1 each collocation is located.

[0070] In addition, the control unit 11 calculates the evaluation value according to the postal code, telephone number, and address. Specifically, the control unit 11 corrects the evaluation value of the collocation according to the positional relationship with the postal code, telephone number, and address (see Figs. 6 to 10). In addition, the control unit 11 calculates the evaluation value according to the position of the collocation (see Fig. 11), calculates the evaluation value corresponding to the keyword "delivery person" of the "delivery note" (see Figs. 12A and 12B), calculates the evaluation value for the keyword "company name" (see Figs. 13A and 13B), calculates the evaluation value corresponding to the keyword "department name" (see Figs. 14A and 14B), calculates the evaluation value corresponding to the keyword "title" (see Figs. 15A and 15B), calculates the evaluation value corresponding to the keyword "transfer destination" (see Fig. 16), and calculates the evaluation value corresponding to the keyword "account" (see Fig. 17).

[0071] The control unit 11 calculates the sum of the evaluation values ​​for each collocation.

[0072] In parallel with step S51, in step S52, the control unit 11 executes a collocation evaluation process. Specifically, the collocation evaluation unit of the control unit 11 calculates an evaluation value based on the characteristics of the collocation itself. Specifically, the control unit 11 calculates an evaluation value corresponding to the font size (see FIG. 18), calculates an evaluation value corresponding to the type of text (national agency, corporation, department, preliminary candidate, etc.) (see FIG. 19), calculates an evaluation value corresponding to the characteristics of the address (see FIG. 21 and FIG. 22), and calculates an evaluation value corresponding to the registration number (see FIG. 23).

[0073] After steps S51 and S52, the control unit 11 shifts the process to step S6. In step S6, the control unit 11 determines the collocation corresponding to the detection target. Specifically, the control unit 11 determines the order of likelihood of each of the issuer, destination, and registration number by comparing the magnitude of the evaluation value obtained by totaling the results of the location information evaluation process (step S51) and the collocation evaluation process (step S52).

[0074] In step S7, the control unit 11 outputs the determination result. For example, the control unit 11 outputs a specific collocation (character string) for each of the issuer, destination, and registration number.

[0075] In this manner, the control unit 11 executes the image processing. The control unit 11 executes the image processing every time an input image is acquired.

[0076] As described above, the image processing device 1 according to this embodiment detects specific character strings (e.g., postal code, address, telephone number) of related items related to a detection target (e.g., issuer, destination, registration number) from character strings to be character-recognized included in an input image (image data P1), calculates an evaluation value for the other character strings based on the position of the specific character string in the image and the positions of other character strings excluding the specific character string, and outputs a character string from the other character strings that is identified based on the evaluation value as the detection target character string. The image processing device 1 also divides an area of ​​the image into a plurality of blocks B1, and calculates the evaluation value based on the positional relationship between a first block corresponding to the position of the specific character string and a second block corresponding to the position of the other character string.

[0077] Normally, when creating a rule between m items, it is a combination calculation, so m × (m-1) calculations are required. In contrast, with the above configuration, the combinations are 2 × m, and the amount of calculations does not increase even if the number of items increases. In particular, when using AI to extract rectangles for OCR words, malfunctions may occur due to moire patterns, etc., and the above configuration is particularly effective in cases where the number of items exceeds 600. In addition, since calculations must be performed using a large number of rules to estimate the destination and issuer while taking into account garbled characters, it is beneficial to reduce the number of combinations. In addition, when setting rules for the positional relationship between keywords and collocations, complex patterns can be realized with a small amount of calculations. In addition, since a large number of rules can be implemented for the same amount of calculations, the accuracy of position estimation is improved.

[0078] Therefore, it is possible to reduce the amount of processing required for recognition of a predetermined item of an input image and to improve the recognition accuracy.

[0079] [Other embodiments] A method for calculating the evaluation values ​​of the "name" and "corporate name" included in the preliminary candidates detected by the detection processing unit 113 will be described below.

[0080] First, the control unit 11 (e.g., the acquisition processing unit 111) acquires the result of OCR processing executed on at least one target character string, for example, a name and / or a corporate name, included in the input image (image data P1). The evaluation processing unit 114 calculates an evaluation value for each of the multiple registered character strings based on a setting value set for each of multiple character string candidates included in the OCR result and a registered value associated with each of multiple registered character strings that are registered in advance and correspond to the target character string. Furthermore, the judgment processing unit 115 (identification processing unit) identifies a specific registered character string based on the evaluation value of each of the multiple registered character strings calculated by the evaluation processing unit 114. Then, the output processing unit 116 outputs the specific registered character string identified by the judgment processing unit 115.

[0081] The evaluation processing unit 114 executes a surname evaluation process to calculate an evaluation value of the surname of the given name, a first name evaluation process to calculate an evaluation value of the first name of the given name, and a corporate name evaluation process to calculate an evaluation value of the corporate name for each collocation included in the preliminary candidates.

[0082] Specifically, the evaluation processing unit 114 calculates an evaluation value for the possibility that the preliminary candidate is a person's surname in the following procedure. The evaluation processing unit 114 calculates an evaluation value for the likelihood that the name is a Japanese name. Since Japanese names are composed of a surname and a given name, the evaluation processing unit 114 calculates the evaluation value of the surname and the evaluation value of the given name separately, and performs a weighted addition of these to calculate an evaluation value for the entire name.

[0083] <Surname evaluation process> FIG. 25 shows an example of the OCR results (first to fifth candidates) of the collocations included in the preliminary candidates.

[0084] First, the evaluation processing unit 114 checks the first or second character of the collocation of the preliminary candidate against the surname table (see FIG. 26). Specifically, the evaluation processing unit 114 checks each of the first to fifth candidates of the OCR result. The surname table shown in FIG. 26 registers currently used surnames and registered values. The registered value is, for example, the number of times the surname is currently used (frequency of use). The surnames in the surname table are selected in order of the number of families, and consist of the top 4000 surnames with one character and two characters, and the top 500 surnames with three characters and four characters. In this way, the evaluation processing unit 114 acquires the registered values ​​by referring to the storage unit (the surname table) in which a predetermined number of registered character strings are registered based on the order of frequency of use.

[0085] In the collation process, the evaluation processing unit 114 assigns a score to each character of each surname registered in the surname table shown in Fig. 26 depending on whether the character is included in the first to fifth candidates of the OCR result, and calculates an evaluation value of the surname based on the score and the registration value registered in the surname table. For example, the evaluation processing unit 114 assigns "5 points" to the first candidate, "4 points" to the second candidate, "3 points" to the third candidate, "2 points" to the fourth candidate, and "1 point" to the fifth candidate.

[0086] For example, as shown in FIG. 27, for the name "Sato" in the surname table, the first character "Sa" and the second character "Fuji" are not included in the first to fifth candidates shown in FIG. 25, so the evaluation processing unit 114 calculates the evaluation value of "Sato" as "0".

[0087] Next, for "Suzuki" in the family name table, the first character "Suzuki" is included in the first candidate and the second character "Ki" is included in the second candidate, so the evaluation processing unit 114 assigns 5 points to "Suzuki" and 4 points to "Ki." In addition, the evaluation processing unit 114 multiplies the total point "9" by the registered value "1778" corresponding to "Suzuki" (see FIG. 26), and calculates the evaluation value of "Suzuki" as "16002."

[0088] In the same manner, the evaluation processing unit 114 calculates an evaluation value for each surname registered in the surname table shown in Fig. 26 (see Fig. 27). After calculating the evaluation value for each surname registered in the surname table, the evaluation processing unit 114 identifies the maximum evaluation value. In this case, the evaluation processing unit 114 identifies the evaluation value "16002".

[0089] The maximum value "16002" obtained above can be regarded as an evaluation value of the surname-likeness. According to the above configuration, when the issuer and the recipient are a person's name (first name), it is possible to obtain the evaluation value for identifying the name from among multiple words written on the document by the above calculation alone. In addition, the addition is performed regardless of the number of characters in the surname, which is convenient because the evaluation value is higher when a longer surname is matched. The number of characters may be weighted by 1 or the number of characters. Even if the OCR reading fails and the candidate is low, the evaluation value decreases as the possibility that the surname is included in the word decreases, so consistency can be achieved. Even in the case of a surname with a small number of households, the surname is often hit with a low candidate, so a low evaluation value is calculated as the surname and an estimate is possible. This allows the evaluation value to be calculated efficiently with a table with a small amount of data, so the evaluation value can be calculated with less memory and calculations than the conventional method.

[0090] As described above, in the surname evaluation process, the evaluation processing unit 114 identifies, for each character included in the surname constituting the name of the registered string (surname table), a character string candidate that includes the character among multiple character string candidates, obtains a setting value (1 point to 5 points) corresponding to the identified character string candidate, and calculates the evaluation value of the surname of the registered string based on the total value of the setting values ​​for each character and the registered value (surname table) associated with the surname. Also, the judgment processing unit 115 (identification processing unit) identifies, among the registered strings, the registered string with the largest evaluation value calculated by multiplying the total value of the setting values ​​by the registered value.

[0091] <Name evaluation process> When the evaluation processing unit 114 detects a phrase that seems to be a family name, it calculates an evaluation value for the characters that follow the family name.

[0092] Specifically, the evaluation processor 114 calculates the evaluation value as "0" if the length of characters following the family name is 0. Also, the evaluation processor 114 calculates the evaluation value as "0" if the length of characters following the family name is more than 4.

[0093] FIG. 29A shows an example of the OCR result (first to fifth candidates) of the characters following the family name. The evaluation processing unit 114 refers to the name table shown in FIG. 28 and sets the registered values ​​corresponding to the first and second characters of the characters (see FIG. 29B). Next, the evaluation processing unit 114 multiplies the registered values ​​shown in FIG. 29B by the magnifications corresponding to the first to fifth candidates. For example, the magnification is set to "5 times" for the first candidate, "4 times" for the second candidate, "3 times" for the third candidate, "2 times" for the fourth candidate, and "1 time" for the fifth candidate. The evaluation processing unit 114 multiplies the registered values ​​of the first and second characters shown in FIG. 29B by the magnifications corresponding to the candidates. FIG. 29C shows the results calculated by multiplying the registered values ​​by the magnifications.

[0094] Next, the evaluation processing unit 114 sums up the results shown in Fig. 29C and divides the sum (here, "104315") by the number of characters ("2") to calculate the evaluation value ("52157.5") of the name "Taro". Note that the evaluation processing unit 114 may multiply the evaluation value by a weight for weighted addition.

[0095] The evaluation processing unit 114 calculates the evaluation value of the name as the sum ("68159.5") of the evaluation value calculated by the surname evaluation process ("16002") and the evaluation value calculated by the given name evaluation process ("52157.5").

[0096] As described above, in the name evaluation process, the evaluation processing unit 114 obtains the registered value associated with each character included in the first name of the character string candidate, and calculates the evaluation value of the first name of the character string candidate based on the registered value for each character and the setting value (1x to 5x) corresponding to the character string candidate. Furthermore, the evaluation processing unit 114 calculates the total value of the evaluation value of the family name and the evaluation value of the first name as the evaluation value of the first name.

[0097] <Corporate name evaluation processing> The evaluation processing unit 114 calculates an evaluation value for the possibility that the preliminary candidate is a corporate name in the following procedure.

[0098] FIG. 30 shows an example of a word table corresponding to corporate names. In the word table, 1024 types of words that represent corporate names and registered values ​​are registered. The registered value is, for example, the current number of times the word is used (frequency of use). For example, the registered value is calculated and registered by multiplying the word length by the number of times it appears in the tax office list. The top 1024 registered values ​​are selected and registered. If the candidates contain part of a word, evaluation values ​​are compared and registered. For example, if "general incorporated foundation" and "foundation incorporated foundation" are included, the respective registered values ​​are compared, the lower one is deleted, and the higher one is registered.

[0099] FIG. 31 shows an example of a character table corresponding to a corporate name. In the character table, characters used in corporate names and registered values ​​are registered. For example, the characters are registered from the names of 5.5 million corporations in a tax office corporate list. The registered values ​​are calculated by simply accumulating the number of times the characters appear in the tax office list. For example, 5,637 types of characters are registered in the character table of FIG. 31.

[0100] FIG. 32A shows an example of the OCR results (first to fifth candidates) of the collocations included in the preliminary candidates.

[0101] The evaluation processing unit 114 compares each character of the preliminary candidate collocation with the word table (see FIG. 26). Specifically, the evaluation processing unit 114 performs comparison for each of the first to fifth candidates of the OCR result.

[0102] In the matching process, the evaluation processing unit 114 performs matching on characters included in the first to fifth candidates among the words registered in the word table shown in Fig. 30, and identifies matching words. The example shown in Fig. 32A shows an example of matching with "West Japan" in the word table. Also, it is assumed that the registration value "57" is registered for "West Japan" in the word table.

[0103] The evaluation processing unit 114 assigns a score to each character in the matched word depending on whether the character is included in the first to fifth candidates in the OCR results, and calculates the evaluation value of the word based on the score and the registered value registered in the word table. For example, the evaluation processing unit 114 assigns "5 points" to the first candidate, "4 points" to the second candidate, "3 points" to the third candidate, "2 points" to the fourth candidate, and "1 point" to the fifth candidate.

[0104] For example, as shown in FIG. 32A, for "West Japan," the first character "Nishi" is included in the second candidate, the second character "Sun" is included in the first candidate, and the third character "Hon" is included in the third candidate, so the evaluation processing unit 114 calculates a total score of "12 points" for "West Japan."

[0105] The evaluation processing unit 114 multiplies the total score "12" by the registered value "57" in the word table for the word "West Japan," to calculate an evaluation value of "684."

[0106] Next, as shown in FIG. 32B, the evaluation processing unit 114 sets a registration value from the character table (see FIG. 31) for each character of the first to fifth candidates of the OCR result. Next, the evaluation processing unit 114 multiplies the registration value shown in FIG. 32B by a magnification corresponding to each of the first to fifth candidates. For example, the magnification is set to "5 times" for the first candidate, "4 times" for the second candidate, "3 times" for the third candidate, "2 times" for the fourth candidate, and "1 time" for the fifth candidate. The evaluation processing unit 114 multiplies the registration value of each character shown in FIG. 32B by the magnification corresponding to the candidate. FIG. 32C shows a result calculated by multiplying the registration value by the magnification.

[0107] Next, the evaluation processing unit 114 sums up the results shown in Fig. 32C and divides the sum (here, "54657") by the average number of characters (for example, "7") to calculate the character evaluation value ("7808.143"). The evaluation processing unit 114 may multiply the evaluation value by a weight for weighted addition.

[0108] The evaluation processing unit 114 calculates the evaluation value of the corporate name as the sum ("8492.143") of the evaluation value of the word ("684") and the evaluation value of the character ("7808.143"). In this way, the evaluation processing unit 114 calculates the evaluation values ​​for the name and the corporate name.

[0109] In this way, in the corporate name evaluation process, the evaluation processing unit 114 obtains the registered value associated with each character contained in the corporate name of the candidate string, and calculates the evaluation value of the corporate name based on the registered value for each character and the setting value (1x to 5x) corresponding to the candidate string.

[0110] Furthermore, the image processing device 1 detects the issuer or destination included in the form image based on the evaluation value of the name and the evaluation value of the corporate name.

[0111] As described above, the image processing device 1 according to this embodiment obtains the results of character recognition processing performed on at least one target character string of a name and / or corporate name included in an image, calculates an evaluation value for each of a plurality of registered character strings based on a setting value set for each of a plurality of character string candidates included in the result of the character recognition processing and a registered value associated with each of a plurality of pre-registered registered character strings corresponding to the target string, identifies a specific registered string based on the calculated evaluation value for each of the plurality of registered character strings, and outputs the identified specific registered string.

[0112] In addition, the image processing device 1 acquires the registered values ​​by referring to a storage unit (such as a surname table, a given name table, a word table, a character table, etc.) in which a predetermined number of registered character strings are registered based on the order (ranking, etc.) of frequency of use.

[0113] According to the conventional method, a search is performed from 220,000 Japanese surnames. In this embodiment, the surname is limited to 1 to 6 characters, so the memory capacity can be reduced from 220,000 to 4,500, almost in proportion to the number of samples. The amount of calculation is about one tenth of the amount if the number of characters in the surname is 2. In addition, the calculation is completed when the comparisons are sequentially performed and a match is found. In the conventional method, 550,000 comparisons are required, since 10 comparisons of 2 characters x candidates 1 to 5 and one comparison to find the maximum value. In contrast, in this embodiment, other calculations are considered as one comparison, and since there are 4,500 items with 12 comparisons per item, 54,000 comparisons are sufficient.

[0114] The number of data entries for corporate names is 5.5 million (approximately 400,000 when considering the number of businesses that have gone out of business), but in this embodiment, the number of entries can be reduced by registering approximately 7,000 entries. Also, since there is a large variation in corporate names, with an average of 12 characters and a maximum of 137 characters, it is difficult to calculate the amount of reduction, but since the order of reduction in the number of tables is not significantly different from that of names, a similar reduction can be expected.

[0115] Furthermore, according to this embodiment, even for incomplete collocations caused by garbled characters or the like, it is possible to calculate a plausible evaluation value from a portion of the character string contained therein.

[0116] The above-mentioned method of calculating the evaluation values ​​of "name" and "corporate name" can be applied to the process of identifying "name" from an input image and the process of identifying "corporate name" from an input image. For example, the image processing device 1 may be configured to calculate the evaluation value of "name" and identify "name" from an input image, or to calculate the evaluation value of "corporate name" and identify "corporate name" from an input image. In other words, the image processing device 1 may not include the process of detecting the issuer, destination, and registration number from the input image. In this case, the image processing device 1 may omit the process of dividing the area of ​​the input image into a plurality of blocks, the location information evaluation process, and the collocation evaluation process.

[0117] The image processing device 1 is an example of an image processing system according to the present disclosure. That is, the image processing system according to the present disclosure may be configured by the image processing device 1 alone. The image processing system according to the present disclosure may also be configured by the image processing device 1 and a character recognition device that executes OCR processing. The image processing system according to the present disclosure may also be configured by the image processing device 1, the character recognition device, and an output device that outputs (displays) the character recognition result. For example, the output device may be a user terminal (personal computer, smartphone, etc.), and the user terminal may execute a character recognition application (Web application) to request the image processing device 1 to perform character recognition processing of a document image, and obtain and display the character recognition result from the image processing device 1.

[0118] [Disclosure Note 1] The following will provide an overview of the disclosure extracted from the above-described embodiment. Note that the configurations and processing functions described in the following supplementary notes can be selected and combined as desired.

[0119] <Appendix 1> a detection processing unit that detects a specific character string related to a detection target from among character strings included in an image and to be subjected to character recognition; an evaluation processing unit that calculates evaluation values ​​of the other character strings based on a position of the specific character string in the image and positions of the other character strings among the character strings excluding the specific character string; an output processing unit that outputs, from the other character strings, a character string that is identified based on the evaluation value calculated by the evaluation processing unit as the detection target character string; An image processing system comprising:

[0120] <Appendix 2> the detection target is an issuer or a destination included in a form image, The specific character string is at least one of a postal code, an address, and a telephone number. 2. An image processing system as described in claim 1.

[0121] <Appendix 3> the evaluation processing unit divides an area of ​​the form image into a plurality of blocks, and calculates the evaluation value based on a positional relationship between a first block corresponding to a position of the specific character string and a second block corresponding to a position of the other character string; 3. An image processing system as described in claim 2.

[0122] <Appendix 4> the evaluation processing unit calculates the evaluation value based on a total number of the specific character strings included in a predetermined range that is set according to the detection target with the position of the first block as a center. 4. An image processing system as described in claim 3.

[0123] <Appendix 5> the evaluation processing unit calculates the evaluation value of the other character string to be higher the closer the other character string is to the specific character string, 5. The image processing system according to any one of claims 2 to 4.

[0124] <Appendix 6> the evaluation processing unit calculates the evaluation value based on a positional relationship between the first block and the second block, a position of the first block in the image, and a position of a predetermined character string among the character strings in the image. 6. An image processing system according to any one of claims 2 to 5.

[0125] <Appendix 7> the evaluation processing unit further calculates the evaluation value based on characteristics of the other character strings. 7. An image processing system according to any one of claims 2 to 6.

[0126] <Appendix 8> the output processing unit outputs a character string of the issuer and a character string of the destination identified from the image. An image processing system according to any one of appendixes 2 to 7.

[0127] [Disclosure Note 2] The following will provide an outline of the disclosure extracted from the above-described other embodiments. Note that the configurations and processing functions described in the following supplementary notes can be selected and combined as desired.

[0128] <Appendix 1> an acquisition processing unit that acquires a result of a character recognition process executed on at least one target character string of a person's name and / or a corporate name included in an image; an evaluation processing unit that calculates an evaluation value for each of a plurality of registered character strings based on a setting value that is set for each of a plurality of character string candidates included in a result of the character recognition process and a registered value that is associated with each of a plurality of registered character strings that are registered in advance and correspond to the target character string; an identification processing unit that identifies a predetermined registered character string based on the evaluation value of each of the plurality of registered character strings calculated by the evaluation processing unit; an output processing unit that outputs the predetermined registered character string identified by the identification processing unit; An image processing system comprising:

[0129] <Appendix 2> the evaluation processing unit refers to a storage unit in which a predetermined number of registered character strings are registered based on the order of frequency of use, and acquires the registered value; 2. An image processing system as described in claim 1.

[0130] <Appendix 3> the evaluation processing unit, for each character included in a surname constituting the name of the registered character string, identifies a character string candidate including the character among the multiple character string candidates, obtains the setting value corresponding to the identified character string candidate, and calculates the evaluation value of the surname of the registered character string based on a total value of the setting values ​​for each character and the registered value associated with the surname; 3. An image processing system according to claim 1 or 2.

[0131] <Appendix 4> the identification processing unit identifies, from among the registered character strings, the registered character string having the largest evaluation value calculated by multiplying the total of the setting values ​​by the registered value; 4. An image processing system as described in claim 3.

[0132] <Appendix 5> the evaluation processing unit acquires, for each character included in the name constituting the name of the character string candidate, the registered value associated with the character, and calculates the evaluation value of the name of the character string candidate based on the registered value for each character and the setting value corresponding to the character string candidate. 5. The image processing system according to claim 3 or 4.

[0133] <Appendix 6> A sum of the evaluation value of the family name and the evaluation value of the first name is calculated as an evaluation value of the first name. 6. The image processing system according to claim 5.

[0134] <Appendix 7> the evaluation processing unit acquires, for each character included in the corporate name of the character string candidate, the registered value associated with the character, and calculates the evaluation value of the corporate name based on the registered value for each character and the setting value corresponding to the character string candidate. 7. An image processing system according to any one of claims 1 to 6.

[0135] <Appendix 8> detecting an issuer or a destination included in the form image based on the evaluation value of the name and the evaluation value of the corporate name; An image processing system according to any one of claims 1 to 7. [Explanation of symbols]

[0136] 1: Image processing device 11: Control section 12: Storage section 13: Operation display section 14: Communications Department 111: Acquisition processing unit 112: Collocation processing unit 113: Detection processing unit 114: Evaluation processing unit 115: Judgment processing unit 116: Output processing section A1 :Area B1: Block P1: Image data

Claims

1. an acquisition processing unit that acquires a result of a character recognition process executed on at least one target character string of a person's name and / or a corporate name included in an image; an evaluation processing unit that calculates an evaluation value for each of a plurality of registered character strings based on a setting value that is set for each of a plurality of character string candidates included in a result of the character recognition process and a registered value that is associated with each of a plurality of registered character strings that are registered in advance and correspond to the target character string; an identification processing unit that identifies a predetermined registered character string based on the evaluation value of each of the plurality of registered character strings calculated by the evaluation processing unit; an output processing unit that outputs the predetermined registered character string identified by the identification processing unit; An image processing system comprising:

2. the evaluation processing unit refers to a storage unit in which a predetermined number of registered character strings are registered based on the order of frequency of use, and acquires the registered value; The image processing system according to claim 1 .

3. the evaluation processing unit, for each character included in a surname constituting the name of the registered character string, identifies a character string candidate including the character among the multiple character string candidates, obtains the setting value corresponding to the identified character string candidate, and calculates the evaluation value of the surname of the registered character string based on a total value of the setting values ​​for each character and the registered value associated with the surname; The image processing system according to claim 1 .

4. the identification processing unit identifies, from among the registered character strings, the registered character string having the largest evaluation value calculated by multiplying the total of the setting values ​​by the registered value; The image processing system according to claim 3 .

5. the evaluation processing unit acquires, for each character included in the name constituting the name of the character string candidate, the registered value associated with the character, and calculates the evaluation value of the name of the character string candidate based on the registered value for each character and the setting value corresponding to the character string candidate. The image processing system according to claim 3 .

6. A sum of the evaluation value of the family name and the evaluation value of the first name is calculated as an evaluation value of the first name. The image processing system according to claim 5 .

7. the evaluation processing unit acquires, for each character included in the corporate name of the character string candidate, the registered value associated with the character, and calculates the evaluation value of the corporate name based on the registered value for each character and the setting value corresponding to the character string candidate. The image processing system according to claim 1 .

8. detecting an issuer or a destination included in the form image based on the evaluation value of the name and the evaluation value of the corporate name; The image processing system according to claim 1 .

9. Obtaining a result of a character recognition process performed on at least one target character string of a person's name and / or a corporate name included in an image; calculating an evaluation value for each of a plurality of registered character strings based on a setting value that is set for each of a plurality of character string candidates included in a result of the character recognition process and a registered value that is associated with each of a plurality of registered character strings that are registered in advance and correspond to the target character string; identifying a specific registered character string based on the evaluation value of each of the plurality of registered character strings; outputting the predetermined registered character string; An image processing method executed by one or more processors.

10. Obtaining a result of a character recognition process performed on at least one target character string of a person's name and / or a corporate name included in an image; calculating an evaluation value for each of a plurality of registered character strings based on a setting value that is set for each of a plurality of character string candidates included in a result of the character recognition process and a registered value that is associated with each of a plurality of registered character strings that are registered in advance and correspond to the target character string; identifying a specific registered character string based on the evaluation value of each of the plurality of registered character strings; outputting the predetermined registered character string; An image processing program for causing one or more processors to execute the above.

Images