Character candidate proposal apparatus, handwritten character recognition system, method and program product

Abstract

Provided is a technique for more appropriately proposing similar character candidates. A character candidate proposal device includes: a similar character storage unit that stores a similarity relation of appearances of characters; a real character storage unit that stores real existing character strings; a trust processing unit that receives one or more characters and accepts a request to propose character candidates similar in appearance to the characters; a comparison processing unit that specifies candidates of characters similar in appearance to the received characters by comparing with the similar character storage unit, compares combinations of the candidates of characters similar in appearance with the real existing character strings stored in the real character storage unit; and a candidate character transmission processing unit that outputs, as candidate characters, combinations of the candidates of characters similar in appearance that are hit by the comparison by the comparison processing unit.

Description

TECHNICAL FIELD

[0001] The present application relates to a character candidate proposal device, a handwritten character recognition system, a handwritten character recognition method, and a program. The present application claims priority from Japanese Patent Application No. 2019-163641 filed on September 9, 2019, the contents of which are incorporated herein by reference for all purposes to the extent permitted by law in the designated states on the basis of document citation. BACKGROUND

[0002] In Patent Literature 1, it is described that: “An information processing device is provided with: a character extraction section 224 that extracts a character string from image data; a list creation section 244 that creates a conversion candidate list containing one or more conversion candidate characters corresponding to an input character; a designation section 248 that designates a character of a conversion target by comparing the one or more conversion candidate characters contained in the conversion candidate list with the character string extracted by the character extraction section; and a display control section 252 that generates a display screen containing the character of the conversion target designated by the designation section”.

[0003] PRIOR ART DOCUMENTS

[0004] PATENT LITERATURE

[0005] Patent Literature 1: Japanese Patent Application Publication No. 2018-163413 SUMMARY

[0006] PROBLEMS TO BE SOLVED BY THE INVENTION

[0007] The above-described technology can present candidates of characters, but if the character recognition is erroneous, it is difficult to obtain a proper result.

[0008] An object of the present application is to provide a technology that more properly proposes similar character string candidates.

[0009] SOLUTION TO THE PROBLEM

[0010] The present application includes a plurality of schemes for solving at least a part of the above-described problems, but if examples thereof are given, they are as described below. An information processing apparatus according to one aspect of the present application includes: a similar character storage section that stores a similar relationship of appearances of characters; a real character storage section that stores a string of a real character; a trusted processing section that receives one or more characters and accepts a request for a character candidate similar in appearance to the characters; a collation processing section that specifies a candidate of a character similar in appearance to the received characters by collation with the similar character storage section, collates a combination of the candidates of the characters similar in appearance with the string of the real character stored in the real character storage section, and a candidate character transmission processing section that outputs, as a candidate character, a combination of the candidates of the characters similar in appearance that matches the result of the collation by the collation processing section.

[0011] Also, for example, the character candidate proposal apparatus described above can include a similar character registration processing section that, after outputting the candidate character, accepts a case where a character similar to any of the characters received by the trusted processing section, and if the similar relationship is not stored in the similar character storage section, causes the similar character storage section to store the similar relationship.

[0012] Also, for example, in the character candidate proposal apparatus described above, the string of the real character can be a string of a real person's name, the trusted processing section can accept a request for a person's name, and the candidate character transmission processing section can output, as a candidate character of a person's name, a combination of the candidates of the characters similar in appearance.

[0013] Also, for example, in the character candidate proposal apparatus described above, the string of the real person's name can be divided into a family name and a given name, and the character candidate proposal apparatus can include a name discrimination processing section that divides the received characters into a family name and a given name and causes the collation processing section to collate.

[0014] Also, for example, in the character candidate proposal apparatus described above, the name discrimination processing section can specify, for the received characters, a combination of dividing a family name and a given name by changing a position at which the family name and the given name are disconnected, and for each of the combinations of the family name and the given name, output, as the candidate character, a combination that matches the result of the collation by the collation processing section.

[0015] Also, for example, in the character candidate proposal apparatus described above, the candidate character transmission processing section can sort and output the candidate characters in order of frequency of occurrence.

[0016] Also, for example, the character candidate proposal device described above can be characterized in that the character received by the entrusted processing section is a character that has been read and recognized.

[0017] Also, another aspect of the present application relates to a handwritten character recognition system including a character candidate proposal device, a service terminal communicably connected to the character candidate proposal device via a network, and a reading device that reads a handwritten character in response to an instruction from the service terminal, the handwritten character recognition system being characterized in that the service terminal is provided with a character recognition section that obtains the handwritten character from the reading device and performs character recognition, a commissioned processing section that sends a character recognized by the character recognition section to the character candidate proposal device and commissions proposal of a character candidate, and a visual confirmation processing section that displays a character candidate output by the character candidate proposal device, the character candidate proposal device being provided with a similar character storage section that stores a similar relationship of the appearance of a character, a real character storage section that stores a real existing character string, an entrusted processing section that receives one or more characters and accepts commission of proposal of a character candidate similar in appearance to the character, a collation processing section that specifies a candidate of a character similar in appearance to the received character by collation with the similar character storage section, collates a combination of the candidates of the character similar in appearance with the real existing character string stored in the real character storage section, and a candidate character sending processing section that outputs, as a candidate character, a combination of the candidates of the character similar in appearance that collated by the collation processing section.

[0018] Further, another aspect of the present application relates to a handwritten character recognition method by a handwritten character recognition system including a character candidate proposal device, a service terminal communicably connected with the character candidate proposal device through a network, and a reading device that reads a handwritten character in response to an instruction from the service terminal, the handwritten character recognition method characterized by the service terminal executing a character recognition step of obtaining the handwritten character from the reading device and performing character recognition, a commission processing step of sending a character recognized in the character recognition step to the character candidate proposal device and commissioning proposal of a character candidate, and a visual confirmation processing step of displaying a character candidate output from the character candidate proposal device, the character candidate proposal device having a similar character storage section that stores a similar relationship of appearances of characters and a real character storage section that stores a real existing character string, the character candidate proposal device executing a trust processing step of receiving one or more characters and accepting commission of proposal of a character candidate similar in appearance to the character, a collation processing step of specifying a candidate of a character similar in appearance to the received character by collation with the similar character storage section, collating a combination of the candidates of the characters similar in appearance with the real existing character string stored in the real character storage section, and a candidate character sending processing step of outputting, as a candidate character, a combination of the candidates of the characters similar in appearance that hit as a result of collation in the collation processing step.

[0019] Further, another aspect of the present application relates to a program that causes a computer to perform a process of assisting recognition of a handwritten character, the program characterized by the computer having a similar character storage section that stores a similar relationship of appearances of characters, a real character storage section that stores a real existing character string, and a processing section, the program causing the processing section to execute a trust processing step of receiving one or more characters and accepting commission of proposal of a character candidate similar in appearance to the character, a collation processing step of specifying a candidate of a character similar in appearance to the received character by collation with the similar character storage section, collating a combination of the candidates of the characters similar in appearance with the real existing character string stored in the real character storage section, and a candidate character sending processing step of outputting, as a candidate character, a combination of the candidates of the characters similar in appearance that hit as a result of collation in the collation processing step.

[0020] Effects of the Invention

[0021] According to the present application, it is possible to provide a technology capable of properly evaluating the content of an article.

[0022] The above problems, configurations, and effects can be understood from the following embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a block diagram illustrating an outline of a handwriting character recognition system according to the present embodiment.

[0024] Figure 2 is a diagram illustrating an example of the configuration of a handwriting character recognition system.

[0025] Figure 3 is a diagram illustrating an example of the data structure of a received data storage section.

[0026] Figure 4 is a diagram illustrating an example of the data structure of a family name storage section.

[0027] Figure 5 is a diagram illustrating an example of the data structure of a given name storage section.

[0028] Figure 6 is a diagram illustrating an example of the data structure of a similar character storage section.

[0029] Figure 7 is a diagram illustrating an example of the data structure of a candidate character storage section.

[0030] Figure 8 is a diagram illustrating an example of the data structure of a character storage section.

[0031] Figure 9 is a diagram illustrating an example of the data structure of an image storage section.

[0032] Figure 10 is a diagram illustrating an example of the hardware configuration of a character candidate proposal device.

[0033] Figure 11 is a diagram illustrating an example of the overall processing flow of handwriting character recognition.

[0034] Figure 12 is a diagram illustrating an example of the flow of a name division collation process.

[0035] Figure 13 is a diagram illustrating an example of the flow of a candidate character reception process.

[0036] Figure 14 is a diagram illustrating an example of the flow of a similar character registration process.

[0037] Figure 15 is a diagram illustrating an example of a candidate presentation screen.

[0038] Figure 16 FIG. 1 is a diagram showing an example of surname candidate presentation.

[0039] Reference Signs List:

[0040] 1 … handwritten character recognition system, 50 … network, 100 … character candidate proposal device, 110 … storage section, 111 … received data storage section, 112 … real-time character storage section, 113 … surname storage section, 114 … given name storage section, 115 … similar character storage section, 116 … candidate character storage section, 120 … processing section, 121 … entrusted processing section, 122 … name discrimination processing section, 123 … collation processing section, 124 … similar character registration processing section, 125 … candidate character transmission processing section, 130 … communication section, 200 … service terminal, 210 … storage section, 211 … character storage section, 212 … candidate character storage section, 213 … image storage section, 220 … processing section, 221 … character recognition section, 222 … entrusted processing section, 223 … visual confirmation processing section, 224 … similar character report processing section, 230 … communication section, 300 … reading device. DETAILED DESCRIPTION

[0041] Hereinafter, a handwritten character recognition system 1 according to an embodiment of the present application will be described based on the drawings. In the following embodiments, the description is divided into a plurality of sections or a plurality of embodiments for convenience, but these sections or embodiments are not necessarily independent of each other, but are in a relationship of a part or all of one being a modification, a detailed description, or a supplementary description of the other, unless otherwise specifically mentioned.

[0042] In the following embodiments, when a number or the like (including the number of items, numerical values, amounts, ranges, and the like) of an element is mentioned, the number is not limited to the specific number, but can be more than or less than the specific number, unless otherwise specifically mentioned or it is clear from the principle that the number is limited to the specific number.

[0043] In the following embodiments, the constituent elements (including element steps and the like) are not necessarily essential, unless otherwise specifically mentioned or it is clear from the principle that the constituent elements are essential.

[0044] Similarly, in the following embodiments, when the shape, positional relationship, or the like of a constituent element or the like is mentioned, a case where the shape or the like is substantially similar or analogous to the shape or the like is included, unless otherwise specifically mentioned or it is clear from the principle that the shape or the like is not substantially similar or analogous to the shape or the like. The same applies to numerical values and ranges.

[0045] Also, in all the drawings for explaining the embodiments, the same reference signs are attached to the same members in principle, and repeated explanation thereof is omitted.

[0046] Figure 1 is a block diagram exemplifying an outline of a handwriting character recognition system to which the present embodiment is related. For the handwriting character recognition system 1, a user performs utilization by connecting to the character candidate proposal device 100 through a browser or the like by using a business terminal 200, but is not limited thereto, and each function of the character candidate proposal device 100 can be mounted as application software to the business terminal 200.

[0047] In addition, in the case where the business terminal 200 connects the character candidate proposal device 100, connection is made through a LAN (Local Area Network), a WAN (Wide Area Network), the Internet, a mobile phone network, or the like, or a network 50 that is a communication network in which these are combined. The network can also be a VPN (Virtual Private Network) or the like on a wireless communication network such as a mobile phone communication network.

[0048] An application example of the handwriting character recognition system 1 can be exemplified by a case where the handwriting character recognition system 1 is a business system or the like that accompanies processing of a prescribed written document (for example, a written document in which a name is handwritten in a prescribed format, such as an application for insurance, an application for opening of a bank account, or the like).

[0049] At this time, in terms of a user, it is conceivable to write a name in a prescribed column of a prescribed written document by hand, and a large number of written documents are processed as a matter, and a large amount of information processing is performed by computer processing, and thus it is greatly advantageous in terms of efficiency, and thus writing of a name or the like is generally digitized by OCR (Optical Character Recognition).

[0050] Here, character data electronicization technology of paper documents is advancing, but there is a certain rate of misreading, unread, or the like, and there is no technology that can completely recognize all handwriting characters. Thus, in the end, it is confirmed by a person's visual confirmation whether characters are consistent, and such visual work becomes one of the reasons that hinders reduction of the cost of electronicization of character data of documents. Specifically, in the case where a name is read, although a method of collating a database on a name such as a family name and a given name to improve the recognition rate of characters is adopted, there are characters that are difficult to read, characters that are easily confused, or the like, and there is still a cost of visual confirmation of whether misreading occurs.

[0051] That is, in a case where the read character contains a misread, for example, in a case where a character recognition engine such as OCR misreads the correct family name "Kawamoto" as "Ogaki", because the misread "Ogaki" is collated with each database, there is a possibility that the correct "Kawamoto" is ranked low in the candidates or is missed as a candidate.

[0052] Also, in a case where the format is to fill in the full name in a column where the family name part and the given name part are not separately specified, if the correct family name is not recognized, there is a possibility that the misread is not recognized, for example, in a case where "Tian" of the full name "Xiaoshantian Shen" is misread as "Yu", the family name is recognized as "Xiaoshan" and a given name such as "Youshen" starting with "Yu" is proposed as a candidate.

[0053] In view of this, the present application also takes a case where the character read by a character recognition engine such as OCR is incorrect as a premise, in addition to collating with a DB (database) regarding the family name and the given name, a similar character in which information of a character that is misread more often by a character recognition engine such as OCR in the past is recorded is collated. Accordingly, the range of candidates of the read character containing a character that is likely to be misread is appropriately expanded, a character that is ranked low or is likely to be missed as a candidate is also proposed as a candidate, and the selection of the candidate character is performed on the basis of the discrimination processing of the family name and the given name. Also, such a similar character can be accumulated with new similar character information by a visual confirmation work of the character, so that the accuracy of character recognition is improved.

[0054] The business terminal 200 according to the present embodiment is a terminal for a business of digitizing such a prescribed written document. The reading device 300 is a scanner, a camera, or the like, and reads the prescribed written document as an image or the like in response to an instruction from the business terminal 200, and delivers the read handwritten character image to the business terminal 200.

[0055] In the business, the business terminal 200, upon receiving the handwritten character image, converts the handwritten character image into a character string by OCR, and transmits the converted character string to the character candidate proposal device 100 and receives the proposed candidate character from the character candidate proposal device 100. Also, the business terminal 200 selectively displays the candidate character, and utilizes the candidate character selected by the user as digitized information.

[0056] In addition, the reading device 300 is not limited to a scanner, and can be a mobile phone terminal with a camera, a tablet terminal with a camera, or the like, which can read an image as data, and can be configured to be communicably connected to the business terminal 200 through the network 50.

[0057] Also, the reading device 300 is not a necessary device, and the character candidate proposal device 100 or the business terminal 200 can receive the read image data, the data on which character recognition has been performed, from other devices through the network 50 to propose the character candidate.

[0058] Figure 2 is a diagram showing an example of the configuration of the handwritten character recognition system. In the character candidate proposal device 100, the storage section 110, the processing section 120, and the communication section 130 are communicably connected to each other through a bus or the like.

[0059] The storage section 110 includes the received data storage section 111, the real character storage section 112, the similar character storage section 115, and the candidate character storage section 116. The real character storage section 112 stores strings of the names of persons, proper nouns, and the like that actually exist. For example, the real character storage section 112 includes a family name storage section 113 and a given name storage section 114. For example, the family names of Japanese people are said to be more than 300,000, and the family names that are confirmed to actually exist are stored in the storage section 113.

[0060] Figure 3 is a diagram showing an example of the data structure of the received data storage section. The received data storage section 111 includes a trust date 111A, a client enterprise ID 111B, a document number 111C, a character recognition number 111D, and a recognized character code 111E. The trust date 111A is information that specifies the date of the acceptance of the request for the character candidate proposal from the business terminal 200. The client enterprise ID 111B is information that specifies the enterprise that has requested the character candidate proposal. The document number 111C is information that specifies the document (image data of the document) that is the object of the request for the character candidate proposal. The character recognition number 111D is information that specifies each character included in the document that is the object of the recognition. The recognized character code 111E is the character code of the character obtained as a result of the recognition of the appearance of each character.

[0061] Figure 4 is a diagram showing an example of the data structure of the family name storage section. The family name storage section 113 includes a family name discrimination symbol 113A, a family name 113B, a number of characters 113C, and a frequency of occurrence 113D. The family name discrimination symbol 113A is information that discriminates the family name from other family names. The family name 113B is the written information of the family name. The number of characters 113C is the number of characters in the writing of the family name. The frequency of occurrence 113D is the frequency of occurrence in the population of the family name.

[0062] Figure 5Fig. 10 is a diagram showing an example of a data structure of the name storage section. The name storage section 114 includes a name discriminator 114A, a name 114B, a character number 114C, and a frequency of occurrence 114D. The name discriminator 114A is information that discriminates a name from other names. The name 114B is written information of the name. The character number 114C is the number of characters in writing of the name. The frequency of occurrence 114D is the frequency of occurrence in the entire body of the name.

[0063] Figure 6 Fig. 11 is a diagram showing an example of a data structure of the similar character storage section. The similar character storage section 115 includes a character code 115A, a similar character code 115B, a frequency of occurrence 115C, and a storage date 115D. The character code 115A is a code of a character that is a reference among characters in a similar relationship. The similar character code 115B is a code of a character that is similar in appearance to the character designated by the character code 115A. The frequency of occurrence 115C is the frequency of occurrence as a similar relationship in use of the character. The storage date 115D is information that designates a date of storage as a similar relationship.

[0064] Figure 7 Fig. 12 is a diagram showing an example of a data structure of the candidate character storage section. The candidate character storage section 116 includes a character discrimination number 116A and a candidate character code 116B. The character discrimination number 116A is information that designates each character included in a document that is an object of discrimination. The candidate character code 116B is a code of a character candidate that is different from a character recognized by OCR, and is a code of a character that is similar and actually exists.

[0065] The processing section 120 includes a trust processing section 121, a name discrimination processing section 122, a collation processing section 123, a similar character registration processing section 124, and a candidate character transmission processing section 125.

[0066] The trust processing section 121 accepts a proposal of a character candidate from the service terminal 200. In a case where a proposal of a character candidate of a plurality of strings is accepted, the name discrimination processing section 122 specifies a combination of a family name and a given name that is disconnected by changing a division position of the name character by character, and discriminates whether or not the combination is a valid name. The collation processing section 123 collates with the actually existing character storage section 112, and determines whether or not it is a character candidate that actually exists. The similar character registration processing section 124 registers a character that is reported by the service terminal 200 as a similar relationship that is not registered, to the similar character storage section 115. The candidate character transmission processing section 125 outputs a combination of character candidates that are similar in appearance as a result of collation by the collation processing section 123, as a candidate character. Specifically, the candidate character transmission processing section 125 realizes a proposal by outputting (transmitting) a candidate saved in the candidate character storage section 116 to the service terminal 200. The communication section 130 communicates with other devices through the network 50.

[0067] In the service terminal 200, the storage section 210, the processing section 220, and the communication section 230 are communicably connected to each other via a bus or the like.

[0068] The storage section 210 includes a character storage section 211, a candidate character storage section 212, and an image storage section 213.

[0069] Figure 8 Fig. 16 is a diagram showing an example of a data structure of the character storage section. The character storage section 211 includes an image data number 211A, a read date 211B, a character recognition number 211C, and an image 211D.

[0070] The image data number 211A is information that specifies an image of a character string read by the reading device 300. The read date 211B is information that specifies a date on which the reading device 300 read a document. The character recognition number 211C is information that specifies each character included in a document that is a recognition target. The image 211D is information that specifies an image data (a portion of an image read by the reading device 300) of a character specified by the character recognition number 211C alone.

[0071] Figure 9 Fig. 17 is a diagram showing an example of a data structure of the image storage section. The image storage section 213 includes an image data number 213A, a document number 213B, a read date 213C, and an image 213D.

[0072] The image data number 213A is information that specifies an image of a character string read by the reading device 300. The document number 213B is information that specifies a document (image data of a document) that is a commission target of a character candidate proposal. The read date 213C is information that specifies a date on which the reading device 300 read a document. The image 213D is information that specifies an image read by the reading device 300.

[0073] The processing section 220 includes a character recognition section 221, a commission processing section 222, a visual confirmation processing section 223, and a similar character report processing section 224. The character recognition section 221 is a so-called OCR engine that analyzes a handwritten character image obtained from the reading device 300 to recognize a character.

[0074] The commission processing section 222 commissions the character candidate proposal device 100 for a proposal of a candidate character with high accuracy in order to eliminate rework, procedure errors, and the like caused by misreading by the character recognition section 221.

[0075] The visual confirmation processing section 223 performs selective display so as to be able to visually confirm each character candidate output by the character candidate proposal device 100 and accepts selection of any candidate from a user.

[0076] The similar character report processing section 224 accepts input of characters identified by the user as similar characters, and reports to the character candidate proposal device 100 so as to be registered as similar characters.

[0077] The communication section 230 communicates with other devices via the network 50.

[0078] In addition, although not illustrated, the service terminal 200 is provided with an input section that accepts input from a user, and an output section that outputs various output information such as screens, account tables, and the like. For example, the input section accepts various inputs such as various contact inputs such as typing, touch, flick input, voice input, or gaze input, and the like.

[0079] Figure 10 is a diagram illustrating a hardware configuration example of the character candidate proposal device 100. The character candidate proposal device 100 is provided with a hardware configuration realized by a housing of a so-called server device, a workstation, a personal computer, a smartphone, or a tablet terminal. The character candidate proposal device 100 is provided with an arithmetic device 101, a main storage device 102, an auxiliary storage device 103, a communication device 104, and a bus 107 that connects each device. The service terminal 200 is also similarly configured. In addition, the service terminal 200 is also provided with input and output devices such as a touch screen, a keyboard, a microphone, a display, and the like.

[0080] The arithmetic device 101 is an arithmetic device such as a CPU (Central Processing Unit).

[0081] The main storage device 102 is a memory device such as a RAM (Random Access Memory).

[0082] The auxiliary storage device 103 is a non-volatile storage device such as a so-called hard disk (Hard Disk Drive), an SSD (Solid State Drive), or a flash memory, that is capable of storing digital information.

[0083] The communication device 104 is a wired communication device such as a network card, or a wireless communication device.

[0084] The above-mentioned trustee processing section 121, the name discrimination processing section 122, the collation processing section 123, the similar character registration processing section 124, and the candidate character transmission processing section 125 of the character candidate proposal device 100 are realized by a program that causes the arithmetic device 101 to perform processing. The program is stored in the main storage device 102, the auxiliary storage device 103, or a ROM device not illustrated, loaded in the main storage device 102 at the time of execution, and executed by the arithmetic device 101.

[0085] Also, the storage section 110 of the character candidate proposal device 100 is realized by the main storage device 102 and the auxiliary storage device 103. Also, the communication section 130 is realized by the communication device 104. The above is an example of the hardware configuration of the character candidate proposal device 100.

[0086] The configuration of the character candidate proposal device 100 can be classified into more constituent elements according to the processing content. Also, one constituent element can be classified so as to perform more processing.

[0087] Also, each processing section (the entrusted processing section 121, the name discrimination processing section 122, the collation processing section 123, the similar character registration processing section 124, and the candidate character transmission processing section 125) can be constructed by a dedicated hardware (ASIC (Application Specific Integrated Circuit), GPU (Graphics Processing Unit), etc.) that realizes each function. Also, the processing of each processing section can be performed by one hardware or by a plurality of hardware.

[0088] Next, the operation of the handwritten character discrimination system 1 of the present embodiment will be described.

[0089] Figure 11 is a diagram showing an example of the entire processing flow of the handwritten character recognition. The handwritten character recognition processing is started by the instruction of the user from the service terminal 200.

[0090] First, the service terminal 200 requests the reading of the image data of the document to the reading device 300 (step S10). Specifically, the character recognition section 221 instructs the reading device 300 to start the scanning or imaging.

[0091] The reading device 300, accepting the request from the service terminal 200, reads the image of the document and creates the image data to respond to the service terminal 200 (step S11).

[0092] The character recognition section 221 of the service terminal 200 receives the image data from the reading device 300, stores it to the image storage section 213, and performs the character recognition processing with the OCR engine (step S12). Also, the recognized character data is saved to the character storage section 211 (step S13). Alternatively, the OCR engine can be installed to a personal computer terminal, a server terminal, etc. provided on a network to which the service terminal 200 is connectable, and the processing can be performed by the instruction of the character recognition processing from the service terminal 200.

[0093] Further, in the service terminal 200, the identified character is transmitted from the commission processing section 222 to the character candidate proposal device 100, and the proposal processing of the character candidate is commissioned (step 14).

[0094] Further, the accepted processing section 121 of the character candidate proposal device 100 accepts the identified character from the service terminal 200 and saves it to the accepted data storage section 111 (step S15).

[0095] Further, the comparison processing section 123 compares the accepted character data with the real character storage section 112 and the similar character storage section 115 (step S16).

[0096] Further, the comparison processing section 123 selects the candidate character and saves it to the candidate character storage section 116 (step S17).

[0097] Further, the candidate character transmission processing section 125 transmits the candidate character to the service terminal 200 (step S18).

[0098] The commission processing section 222 of the service terminal 200 accepts the candidate character and saves it to the candidate character storage section 212 (step S19).

[0099] Further, the visual confirmation processing section 223 outputs a screen for comparing the candidate character with the image of the image storage section (step S20). More specifically, for example, the screen displays the image of the handwritten character read by the reading device 300 and a plurality of the candidate characters so that the input can be selected.

[0100] In the case where the registration instruction of the similar character is accepted from the user, the similar character reporting processing section 224 accepts the input if there is a character similar to the candidate character in a single character, and responds to the character candidate proposal device 100 (step S21).

[0101] The similar character registration processing section 124 of the character candidate proposal device 100 accepts the similar character, and saves it if it is not saved in the similar character storage section 115 (step S22).

[0102] The above is an example of the entire processing flow of the handwritten character recognition. By the entire processing of the handwritten character recognition, the similar character candidate can be more appropriately proposed.

[0103] Figure 12 is a diagram showing an example of the flow of the name division comparison processing. The name division comparison processing is an example of the processing of the series of steps S16 to S18 in the entire processing flow of the handwritten character recognition.

[0104] First, the name discrimination processing section 122 specifies the total number of characters n (step S1601). For example, in the case where the name "Shin Koyama" is the string of the object presented as a candidate, the name discrimination processing section 122 specifies the total number of characters n as "4" by regarding each Chinese character "Shin", "Koyama", "Shin", and "Koyama" as one character.

[0105] Further, the name discrimination processing section 122 performs the division of the characters from the head d (d is 1 or more and less than n, and the initial value is 1) as the family name, and the remaining characters as the given name (step S1602).

[0106] The collation processing section 123 performs the collation of the characters constituting the family name one by one with the similar character storage section 115, and collates the combination of the similar characters with the family name storage section 113 to select the actually existing family name candidate (step S1603).

[0107] Further, the name discrimination processing section 122 determines whether there is no family name candidate that matches in the similar character storage section 115 (step S1604). In the case where there is no family name candidate (in the case where "Yes" of step S1604), the name discrimination processing section 122 advances the control to the step S1611 described later.

[0108] In the case where there is a family name candidate (in the case where "No" of step S1604), the name discrimination processing section 122 determines whether there is a family name that matches even one of the family name candidates in the family name storage section 113 (step S1605). In the case where there is no family name that matches in the family name storage section 113 (in the case where "No" of step S1605), the name discrimination processing section 122 advances the control to the step S1611 described later.

[0109] In the case where there is a family name that matches the family name storage section 113 in the family name candidate (in the case where "Yes" of step S1605), the name discrimination processing section 122 temporarily holds all the family name candidates selected (step S1606).

[0110] Further, the collation processing section 123 performs the collation of the characters constituting the given name one by one with the similar character storage section 115, and collates the combination of the similar characters with the given name storage section 114 to select the actually existing given name candidate (step S1607).

[0111] Further, the name discrimination processing section 122 determines whether there is no given name candidate that matches in the similar character storage section 115 (step S1608). In the case where there is no given name candidate (in the case where "Yes" of step S1608), the name discrimination processing section 122 advances the control to the step S1611 described later.

[0112] In the case of a name candidate (the case of "No" in step S1608), the name discrimination processing section 122 determines whether there is a name in the name storage section 114 that matches even one of the name candidates (step S1609). In the case where there is no matching name in the name storage section 114 (the case of "No" in step S1609), the name discrimination processing section 122 advances the control to step S1611 described later.

[0113] In the case where there is a name in the name storage section 114 that matches the name candidate (the case of "Yes" in step S1609), the name discrimination processing section 122 temporarily stores all the selected name candidates (step S1610).

[0114] Further, the name discrimination processing section 122 increments the character number d of the family name by 1 (step S1611).

[0115] Further, the name discrimination processing section 122 determines whether the character number d of the family name coincides with the total character number n (step S1612). In the case of non-coincidence (the case of "No" in step S1612), the name discrimination processing section 122 returns the control to step S1602.

[0116] In the case where the character number d of the family name coincides with the total character number n (the case of "Yes" in step S1612), the name discrimination processing section 122 determines whether there is a set of the temporarily stored family name candidate and name candidate (the family name length + the name length coincides with the total character number n), that is, determines whether there is a combination of names that is a result hit (step S1613).

[0117] In the case where there is no set of the temporarily stored family name candidate and name candidate (the family name length + the name length coincides with the total character number n) (the case of "No" in step S1613), the candidate character transmission processing section 125 notifies the service terminal 200 of the inability to discriminate (step S1614).

[0118] In the case where there is a set of the temporarily stored family name candidate and name candidate (the family name length + the name length coincides with the total character number n) (the case of "Yes" in step S1613), the candidate character transmission processing section 125 transmits the candidate characters to the service terminal 200 (step S1801). At this time, the candidate character transmission processing section 125 sorts and transmits the set of the temporarily stored family name candidate and name candidate in order of the frequency of occurrence for each. That is, the candidate character transmission processing section 125 transmits the set of the family name candidate and name candidate in order of the character having a high frequency of occurrence.

[0119] The above is an example of the flow of the name division reference processing. With the name division reference processing, for a name composed of one or more characters, the division position of the name is changed and respective similar characters are applied, and an actually existing candidate is designated. For example, in the case of the name "Nobuyasu Konoyama" described above, "Konoyama" and "Nobuyasu" are similar characters, and it is possible to designate candidate characters including a candidate in which the two characters of the family name "Konoyama" and the given name "Nobuyasu" are included, and a candidate in which the three characters of the family name "Konoyama" and the given name "Nobuyasu" are included.

[0120] Figure 13 Fig. 16 is an example diagram showing the flow of the candidate character reception processing. The candidate character reception processing is an example of processing in which steps S19 to S21 of the entire processing flow of the handwriting character recognition are performed in succession.

[0121] First, the commission processing section 222 causes the received candidate characters to be stored in the candidate character storage section 212 (step S191).

[0122] Further, the visual confirmation processing section 223 performs a confirmation display of the candidate characters and the handwriting character image (step S201). An example of the screen of this confirmation display is described later.

[0123] Further, the visual confirmation processing section 223 accepts a selection input of the candidate characters (step S202). In accordance with this, the digitization of the handwriting character is determined.

[0124] Further, the similar character report processing section 224 determines whether there is a new similar character input (step S211). In the case where there is no new similar character input (in the case where "No" of step S211), the similar character report processing section 224 ends the candidate character reception processing.

[0125] In the case where there is a new similar character input (in the case where "Yes" of step S211), the similar character report processing section 224 transmits information that becomes the new similar character to the character candidate proposal device 100 (step S212).

[0126] The above is an example of the flow of the candidate character reception processing. With the candidate character reception processing, it is possible to receive the candidate characters from the character candidate proposal device 100, perform a selection input of the candidates while comparing with the handwriting character. Further, in the case where a similar character is found, it is possible to report to the character candidate proposal device 100 and commission registration as a new similar character.

[0127] Figure 14is a drawing showing an example of a flow of similar character registration processing. In a case where information of a similar character that becomes a candidate character in the candidate character reception processing is sent from the service terminal 200 to the character candidate proposal device 100, the similar character registration processing is started by the character candidate proposal device 100.

[0128] First, the similar character registration processing section 124 determines whether or not the received information of a similar character is already stored in the similar character storage section 115 (step S2201).

[0129] In a case where the received information of a similar character is not stored in the similar character storage section 115 (in a case where "No" in step S2201), the similar character registration processing section 124 specifies a combination of similar characters one by one (divides characters in a similar relationship into one-to-one combinations) (step S2202).

[0130] Also, the similar character registration processing section 124 stores each of the characters that are similar to each other in the similar character storage section 115 as similar (step S2203).

[0131] In a case where the received information of a similar character is already stored in the similar character storage section 115 (in a case where "Yes" in step S2201), the similar character registration processing section 124 notifies the service terminal 200 of the already registered information (step S2204).

[0132] Also, the similar character registration processing section 124 increases the frequency of appearance of each of the characters that are similar to each other by a predetermined amount (step S2205).

[0133] The above is an example of a flow of similar character registration processing. Through the similar character registration processing, a similar character that is newly registered by the service terminal 200 can be registered as similar, and the frequency of appearance can be increased in a case where the similar character is already registered.

[0134] Figure 15 is a drawing showing an example of a candidate presentation screen. The candidate presentation screen 400 is a screen output in step S20 of the entire handwriting character recognition processing flow, step S201 of the candidate character reception processing. The original data (a handwriting character image read by the reading device 300) for visual confirmation is displayed in an original data display area 401, and candidate characters that are objects of visual confirmation are displayed in a family name candidate display area 402 and a given name candidate display area 403. The character candidate proposal device 100 proposes candidate characters with the highest frequency of appearance in the family name candidate display area 402 and the given name candidate display area 403. A family name candidate decision button 404, when input, causes the family name displayed in the family name candidate display area 402 to be determined. A given name candidate decision button 405, when input, causes the given name displayed in the given name candidate display area 403 to be determined.

[0135] The next candidate characters are displayed in the next candidate display area 406 for the family name and the next candidate display area 407 for the given name, respectively, in order of frequency of occurrence and are selectable. The next candidate display area 406 for the family name and the next candidate display area 407 for the given name accept selection of characters from the list. The next candidate decision button 408, when input, causes the selected family name in the next candidate display area 406 for the family name to be determined. The next candidate decision button 409, when input, causes the selected given name in the next candidate display area 407 for the given name to be determined.

[0136] The similar character registration area 410 and the similar character registration area 411 accept input of characters, respectively. The registration button 412, when input, causes the characters input to the similar character registration area 410 and the similar character registration area 411 to be accepted as characters in a similar relationship and causes the registration process to the character candidate proposal device 100 to be started. That is, in the case where the visual confirmation result is that the visual work supervisor (user) feels that it is necessary to register a character as a new similar character, the character can be registered by inputting the character to the similar character registration area 410 and the similar character registration area 411 and selecting the registration button 412. Further, as needed, input to the back button 413 or the next screen button 414 is accepted to move to the previous or next screen.

[0137] Figure 16 Fig. 6 is a diagram showing an example of the family name candidate display. The handwriting character recognition system 1 according to the present embodiment, even in the case where the reading device 300 and the character recognition section 221 misread, has a high probability that the character of the correct answer is included in the displayed candidates, and this is described in detail.

[0138] In the case where the character recognition engine such as OCR misreads the family name "Kawamoto" (a) of the correct answer as "Koki" (b), if a comparison is made between the misread "Koki" (b) and the actually existing family names, there is a possibility that the character candidate of the correct answer "Kawamoto" (a) is ranked low or is missed.

[0139] The misread "Koki" (b) is decomposed into "Ko" (c) and "Ki" (d), and a comparison is made with the similar character storage section 115, respectively, to specify similar characters (e, f) in order of high frequency of occurrence. Next, a comparison is made with the family name storage section 113 by combining the characters of high frequency of occurrence with each other to be a family name, and the candidate character (g) is selected and ranked. According to this, even in the case where there is a misreading, by performing a comparison with the similar character storage section 115, it is possible to avoid the correct answer "Kawamoto" (a) being ranked low or being missed, and it is possible to expect an improvement in the character recognition rate.

[0140] The above is an example of a handwritten character recognition system according to the embodiment. With the example of the handwritten character recognition system according to the embodiment, a similar string candidate can be more appropriately proposed for an image in which a handwritten character is read, so that a user can smoothly perform a handwritten character reading service by selecting from the presented candidates.

[0141] Also, in the example of the above embodiment, a name is cited as an example of an object of a handwritten character, but is not limited thereto, and by replacing the real character storage section 112 with which a comparison is made, it can also be applied to all other character data such as an address.

[0142] Also, the technical elements of the above embodiment can be applied individually, or can be applied in multiple parts such as program components and hardware components.

[0143] The above has been described with respect to the present application, centering on an embodiment.

Claims

1. A character candidate proposal device characterized by comprising: a similar character storage section that stores a similar relation of appearances of characters; a real character storage section that stores strings of names of persons that actually exist and are divided into family names and given names; a trust processing section that receives one or more characters and accepts a trust to propose character candidates of names of persons that are similar in appearance to the characters; a comparison processing section that specifies candidates of characters of names of persons that are similar in appearance to the received characters by comparison with the similar character storage section, compares combinations of the candidates of characters of names of persons that are similar in appearance with the strings of names of persons that actually exist stored in the real character storage section; a candidate character transmission processing section that outputs, as a candidate of a name of person, a combination of the candidates of characters of names of persons that are similar in appearance that hit as a result of comparison by the comparison processing section; a name discrimination processing section that divides the received characters into family names and given names and causes the comparison processing section to compare, and the name discrimination processing section specifies, with respect to the received characters, combinations of dividing into family names and given names by changing a position at which the family name and the given name are disconnected, and with respect to both the family name and the given name of each combination, specifies, as the candidate characters, in a case where a hit is obtained as a result of comparison by the comparison processing section, the candidate character transmission processing section sorts and outputs the candidate characters in order of frequency of appearance.

2. The character candidate proposal device according to claim 1, characterized in that the character candidate proposal device comprises a similar character registration processing section that, after the candidate characters are output, accepts a character that is similar to any one of the characters received by the trust processing section, and if the similar relation is not stored in the similar character storage section, causes the similar character storage section to store the similar relation.

3. The character candidate proposal device according to claim 1, characterized in that the characters received by the trust processing section are characters that are read as handwritten characters and are recognized. a character candidate proposal device; a service terminal communicably connected to the character candidate proposal device through a network; and a reading device that reads a handwritten character in response to an instruction from the service terminal, the handwritten character recognition system characterized in that 4. A system for recognizing handwritten characters, the system comprising: the service terminal comprises: a character recognition section that obtains the handwritten character from the reading device and performs character recognition; a trust processing section that transmits the character recognized by the character recognition section to the character candidate proposal device and trusts a proposal of character candidates; and a visual confirmation processing section that displays the character candidates output by the character candidate proposal device, the character candidate proposal device comprises: a similar character storage section that stores a similar relation of appearances of characters; a real character storage section that stores strings of names of persons that actually exist and are divided into family names and given names; a trust processing section that receives one or more characters and accepts a trust to propose character candidates of names of persons that are similar in appearance to the characters; a comparison processing section that specifies candidates of characters of names of persons that are similar in appearance to the received characters by comparison with the similar character storage section, compares combinations of the candidates of characters of names of persons that are similar in appearance with the strings of names of persons that actually exist stored in the real character storage section; ​ a collation processing section that specifies candidates of characters of a person's name similar in appearance to the received characters by collation with the similar character storage section, collates combinations of the candidates of characters of the person's name similar in appearance with the strings of the actually existing person's name stored in the actually existing character storage section; a candidate character transmission processing section that outputs, as candidates of a person's name, combinations of the candidates of characters of the person's name similar in appearance that hit in the collation by the collation processing section; and a name discrimination processing section that divides the received characters into a family name and a given name and causes the collation processing section to collate, the name discrimination processing section specifies, with respect to the received characters, combinations of a family name and a given name divided by changing the position at which the family name and the given name are separated, and with respect to both the family name and the given name of each combination, the combination is the candidate character in the case where it hits in the collation by the collation processing section, the candidate character transmission processing section sorts and outputs the candidate characters in order of frequency of appearance.

5. A method of recognizing a handwritten character, the method of recognizing a handwritten character being performed by a handwritten character recognition system, the handwritten character recognition system comprising: character candidate proposal device; a service terminal communicably connected to the character candidate proposal device via a network; and a reading device that reads a handwritten character in response to an instruction from the service terminal, the handwritten character recognition method characterized in that the service terminal executes the following steps: a character recognition step in which the handwritten character is obtained from the reading device and character recognition is performed; a commission processing step in which the character recognized in the character recognition step is transmitted to the character candidate proposal device and commission of proposal of a character candidate is requested; and a visual confirmation processing step in which the character candidate output by the character candidate proposal device is displayed, the character candidate proposal device has: a similar character storage section that stores similar relationships of characters in appearance; and an actually existing character storage section that stores strings of actually existing person's names divided into a family name and a given name, the character candidate proposal device executes the following steps: a commission accepting processing step in which one or more characters are received and commission of proposal of a character candidate of a person's name similar in appearance to the character is accepted; a collation processing step in which candidates of characters of a person's name similar in appearance to the received characters are specified by collation with the similar character storage section, combinations of the candidates of characters of the person's name similar in appearance are collated with the strings of the actually existing person's name stored in the actually existing character storage section; and a candidate character transmission processing step in which, as candidates of a person's name, combinations of the candidates of characters of the person's name similar in appearance that hit in the collation in the collation processing step are output, in the collation processing step, the received characters are divided into a family name and a given name and collated, with respect to the received characters, combinations of a family name and a given name divided by changing the position at which the family name and the given name are separated are specified, and with respect to both the family name and the given name of each combination, the combination is the candidate character in the case where it hits in the collation in the collation processing step, The candidate characters are sorted by frequency of occurrence and output in the candidate character transmission processing step.

6. A program product in which a program causes a computer to perform processing for supporting recognition of a handwritten character, the program product characterized by, the computer having: a similar character storage section that stores a similar relationship of appearances of characters; a real character storage section that stores strings of names of persons that actually exist and are divided into family names and given names; and a processing section, the program causing the processing section to perform the following steps: a trusted processing step in which one or more characters are received and a request is accepted for a character candidate of a name of a person whose appearance is similar to the character; a comparison processing step in which a candidate of a name of a person whose appearance is similar to the character received is specified by comparison with the similar character storage section, and combinations of the candidate of the name of the person whose appearance is similar are compared with the strings of names of persons that actually exist and are stored in the real character storage section; a candidate character transmission processing step in which combinations of the candidate of the name of the person whose appearance is similar that hit as a result of the comparison in the comparison processing step are output as candidate characters of a name of a person, in the comparison processing step, the character received is divided into a family name and a given name to perform the comparison, for each combination of a family name and a given name of the character received, a combination in which the family name and the given name are divided is specified by changing a position at which the family name and the given name are disconnected, and the candidate characters are obtained as a result of the comparison in the comparison processing step when a hit is obtained, the candidate characters are sorted by frequency of occurrence and output in the candidate character transmission processing step.

Images