JP6582464B2 - Information input device and program - Google Patents

Description

  The present invention relates to an information input device using optical character recognition that supports an input operation by a user.

  In the past, when a user applies for participation or membership in a membership campaign such as a point card member in a business campaign on the street or in a general commercial transaction, it is described in the personal authentication medium possessed by the user. A method of reducing the burden of filling in an application form for a user by photographing personal information and performing optical character recognition (OCR) has been performed.

  In this case, if the target product to which the campaign or privilege is granted is subject to legal age restrictions on its use, such as alcoholic beverages and tobacco, the personal authentication medium is, for example, a user's license. It is desirable to have a public ID with a date of birth. With such a medium, age authentication can be performed at the same time as application for membership, so the burden on age authentication is reduced for both users and companies.

  In these official IDs such as licenses, the position and character size where the first name, last name, address, date of birth, etc., to be read by the information input device are almost fixed, so in particular, OCR It is considered suitable for the automatic character recognition process used.

  In Patent Document 1, in order to authenticate a procedural member such as a contractor's family in a procedure for changing the contents of a contract in the insurance industry, the surname of the procedural member is obtained from image data of an identification card such as a procedural license. Is obtained as text data, extracted as official identification data, and presented with an authentication device that performs authentication by matching with the contractor's last name.

  Further, Patent Document 2 introduces a character recognition processing device used for recognizing first and last names using an optical character reader (OCR). According to this, when recognizing first and last names, the character string consisting of the first and last names is read, a partition specific to the first and last names is set according to the number of characters, and divided into two parts, the first half and the second half. A method of recognizing a surname and a first name is presented by comparing with a dictionary and comparing the latter half with a name dictionary, and when it cannot be recognized, changing the partition and comparing with each dictionary again.

  Moreover, in patent document 3, in the character recognition method in the optical character reader (OCR) for recognizing first and last names of handwritten Chinese characters (which may be printed characters) written by hand, A method for displaying candidate data for selecting a separation method is introduced.

Japanese Patent No. 5166330 Japanese Patent No. 2892376 Japanese Patent No. 2933178

  By the way, when automatic character recognition is performed using OCR, even if the description position and character size (character height) are aligned, for example, in the first name and last name written in the first and last name field, it is clear between the first name and the last name. The width of the character when there is a large number of characters written due to restrictions on the size of the entry field in the address, etc. May be half-width or 2 / 3-corner of the vertical size, which may make optical character recognition difficult.

  In the method disclosed in Patent Document 1, since the surname of a procedure person to be extracted is known in advance before the last name is determined (it is expected to match the surname of the contractor), the surname It is easy to accurately distinguish the first name and the last name, but in normal character recognition, the last name is not known in advance. In Patent Document 2, as pointed out in Patent Document 3, in the case where there are a plurality of different surname and first name delimiters, any one delimiter is selected. There is a problem that there is no guarantee of being separated. Further, in Patent Document 3, when there are a plurality of different last name and first name separation methods, candidate data for selecting the separation method is displayed, and the user selects a desired first and last name separation from the candidate data. Therefore, the first name and the last name are separated accurately. However, since multiple candidate last name and first name separation methods are displayed, the first name and last name separation method that the user feels uncomfortable is displayed, or an input device such as a keyboard or a mouse is separately displayed from the displayed list of separation methods. Since it is necessary to make a selection by using a space, a physically fixed space or the like is required, which may be bothersome for the user and the company.

  An object of the present invention is to capture personal information described in a medium for personal authentication possessed by a user, accurately extract description data from the image data by optical character recognition (OCR), It is an object to provide an information input device capable of accurately distinguishing a first name and a last name by an easy operation without choosing a place even when there is no clear separation (blank) or the like.

In order to solve the above problems, the first invention of the present application is a photographing means for photographing a public identification card and storing it as image data, and storing the stored image data for each type of the public identification card. Image dividing means for dividing the partial image data into a plurality of partial image data using the determined layout information, image correcting means for scaling the partial image data to a predetermined size, and each part scaled to the predetermined size Character recognition means that converts image data into text data using optical character recognition and retains it together with attributes obtained from layout information, and first and last names that categorize character strings obtained by converting partial image data related to first and last names into text data Classifying means and classifying means for detecting the input indicating the first name and last name separator of the first and last names and determining the text data as the last name text data and first name text data separator Data storing means for storing each text data, and the image correcting means compares the number of characters of the first and last names included in the partial image data relating to the first and last names with the character width of one character in the partial image data relating to the address. Thus, the information input device is characterized in that each character width is determined and each partial image data is enlarged / reduced to have the same character width. In this way, the user's personal information can be acquired by photographing the public identification card possessed by the user, and the accuracy of character recognition by optical character recognition (OCR) is improved by performing correction by enlargement / reduction of characters in advance. Can be achieved.

According to a second aspect of the present invention, there is provided a photographing means for photographing a public identification card and storing it as image data, and a plurality of pieces of the stored image data using layout information determined for each type of the public identification card. Image dividing means for dividing the image into partial image data; image correcting means for enlarging / reducing the partial image data to a predetermined size; and text data for each partial image data enlarged / reduced to the predetermined size using optical character recognition Character recognition means for storing data in combination with attributes obtained from layout information, first-name and last-name classification means for classifying a character string obtained by converting partial image data relating to first and last names into text data, first-name and last-name Classifying means for detecting the input indicating the delimiter and determining the delimiter between the text data for the last name and the text data for the first name and the text data for each. The first and last name classifying means classifies a character string obtained by converting the partial image data relating to the first and last name into text data into one or more last names and first names, and provides an index of the certainty of each separation method. The information input device is characterized in that the classification determination means modifies the character string based on the index and displays it on a display unit. In this way, the user's personal information can be obtained by photographing the public identification card possessed by the user, and the user can more accurately specify the correct first and last name separators.

In a third aspect of the invention, the image correction unit determines the character width by comparing the number of characters of the first and last names included in the partial image data relating to the first and last names and the character width of one character in the partial image data relating to the address. In the information input device according to the second aspect of the present invention, each partial image data is enlarged or reduced so as to have the same character width. Thus, by performing the correction by the enlargement / reduction of characters in advance, the accuracy of character recognition by optical character recognition (OCR) can be improved.

  A fourth invention is a computer program for causing a computer to function as the information input device according to any one of the first to third inventions.

  According to the present invention, it is possible to extract data described on a medium accurately and to easily operate without selecting a place only by photographing personal information described on a medium for personal authentication possessed by a user. The first name and last name can be accurately distinguished.

It is a figure which shows the structure of this embodiment. 2 is a functional block diagram for explaining the information input device 100 in more detail. FIG. 2 is a diagram illustrating a hardware configuration of an information input device 100. FIG. It is a figure explaining the layout information 131. FIG. It is a figure explaining the full name dictionary 161 and the full name history 162. FIG. It is a flowchart explaining the flow of a process of this embodiment. 3 is a diagram illustrating an example of the appearance of a personal authentication medium 200. FIG. 5 is a flowchart relating to correction of character size in an image correction unit 140. 5 is a flowchart for obtaining an index for separating first and last names in first and last name classification means 160. It is a figure which shows the example of the text of the full name which the division | segmentation determination means 170 displays on the display part 105. FIG.

  Hereinafter, the configuration of a system according to an embodiment of the present invention will be described in more detail with reference to the drawings.

  FIG. 1 is a diagram showing a configuration of the present embodiment. Reference numeral 100 denotes an information input device, which is typically a portable terminal such as a tablet computer, and includes a camera 121. In addition, the information input apparatus 100 photographs the medium 200 for personal authentication such as a public identification card possessed by the user with the camera 121. The information input device 100 transmits and receives information to and from an external device via the network 500. Here, in particular, information is exchanged with a member server 300 of a company that manages user personal information.

  FIG. 2 is a functional block diagram for explaining the information input device 100 of the present embodiment in more detail from the aspect of its function. The information input device 100 includes a photographing unit 120 that stores image data photographed by a built-in camera, an image dividing unit 130 that divides the image data into a plurality of partial image data using predetermined layout information, and partial image data. An image correcting unit 140 for enlarging / reducing to a predetermined size, a character recognizing unit 150 for converting each partial image data enlarged / reduced to a predetermined size into text data using optical character recognition, and a surname / name converted to text data are classified The first and last name classification means 160 for displaying the index obtained in that case together with the first and last character strings on the display unit, and the input indicating the last name and first name separator of the displayed first and last names is detected and the text data is converted into the last name text data. Classification determining means 170 for classifying the text data into names, and data storage means 180 for storing the respective text data, respectively. Control means 110 is provided for controlling the means. The process flow will be described in further detail later.

  FIG. 3 is a diagram illustrating a hardware configuration of the information input device 100. As described above, the information input device 100 is typically a tablet computer (hereinafter referred to as a tablet PC), and is a single computer system as hardware. The information input device 100 includes a control unit 101, a storage unit 102, a peripheral device interface (I / F) unit 103, an input unit 104, a display unit 105, and a communication unit 106, which are connected via a bus 109. Further, as a feature of this configuration, a camera 121 connected to the peripheral device I / F unit 103 is incorporated. Note that the hardware configuration in FIG. 3 is merely an example, and various configurations can be adopted depending on purposes.

  The control unit 101 includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The CPU calls a program stored in the storage unit 102, ROM, recording medium or the like to a work memory area on the RAM, executes it, drives and controls each device connected via the bus 109, and performs processing performed by the computer. Realize. The ROM is a non-volatile memory and permanently holds a boot program, data, and the like of the information input device 100. The RAM is a volatile memory, and temporarily stores a program, data, and the like loaded from the storage unit 102, ROM, recording medium, and the like, and includes a work area used by the control unit 101 for performing various processes.

  The storage unit 102 stores a program executed by the control unit 101, data necessary for program execution, an OS (operating system), and the like. Regarding programs, a control program corresponding to an OS (operating system), a program code functioning as the control unit 110, a program code functioning as the photographing unit 120, a program code functioning as the image dividing unit 130, and a function as the image correction unit 140 A program code that functions as the character recognition means 150, a program code that functions as the first and last name classification means 160, a program code that functions as the classification determination means 170, and a program code that functions as the data storage means 180. These program codes are read by the control unit 101 as necessary, transferred to the RAM, and read and executed by the CPU to function as various means.

  The peripheral device I / F unit 103 is a port for connecting a peripheral device to the computer, and the computer transmits and receives data to and from the peripheral device via the peripheral device I / F unit 103. The peripheral device I / F unit 103 is configured by USB, IEEE 1394, RS-232C, or the like. In this embodiment, the peripheral device I / F unit 103 is connected to the camera 121 built in the tablet PC and receives captured data. The input unit 104 is an input device such as a touch panel, for example. Through the input unit 104, the user can perform an operation instruction, an operation instruction, data input, and the like with respect to the information input device 100. The display unit 105 is a display device such as a liquid crystal panel, and may be installed so as to overlap with the touch panel of the input unit 104.

  The communication unit 106 includes a communication control device, a communication port, and the like, and is a wireless communication interface that mediates communication between the information input device 100 and the network 500, and controls communication with other computers via the network 500. Do.

  Returning to FIG. 2, the control unit 110 is a computer program that executes the control unit 101 and an OS (operating system), and the photographing unit 120 is a computer program that controls photographing with the camera. Further, the image dividing unit 130 to the data storage unit 180 are also computer programs related to the recognition of captured images, and function by the control unit 110 interpreting and executing each computer program.

  Below, the database which the information input device 100 of this embodiment has is demonstrated. Further, in this embodiment, these various databases only illustrate the minimum necessary items in the description, but in actual operation, various items derived from these may be added and stored. Good.

  FIG. 4 is a diagram for explaining layout information 131 which is one of databases included in the information input device 100. The image dividing unit 130 of the information input device 100 divides the image data stored by the photographing unit 120 into a plurality of partial image data using the layout information 131.

  The layout information 131 is used in units of personal authentication medium names (T12), and this record constitutes one record. Which personal authentication medium the image dividing unit 130 uses may be set in advance by the operator of the information input apparatus 100 or may be automatically set, as will be described later. The number T11 is an identifier (ID) for the personal authentication medium name (T12).

  For example, when the personal authentication medium is set as a car driver's license, this record includes several attributes (also referred to as items, T13). For example, first name and last name data (one of T13) is used as a reference of the input image. Surrounded by start point coordinates (T14, T15, X = 29, Y = 15 in this example) and end point coordinates (T16, T17, X = 40, Y = 105 in this example) from the point (usually the upper left corner point) Exists in the range. In addition, it is shown that the font of the first and last name characters (font, T18) is “Maru Gothic”, and the vertical size (T19) of the characters is 12 points. The coordinate system, font name, character size unit, and the like may be arbitrarily set in the program.

  FIG. 5 is a diagram for explaining the first name surname dictionary 161 and the first name surname history 162 which are one of the databases of the information input device 100. The first and last name classifying means 160 of the information input device 100 classifies the first name and last name converted into text data by the character recognition means 150 using the first name and first name dictionary 161 and the first name and first name history 162, and uses a first name and last name as an index indicating the likelihood of the classification. Extracted from the dictionary 161 and the first and last name history 162.

  FIG. 5A is an example showing the configuration of the first name surname dictionary 161. In this example, a list of surnames that should exist (T21) and an index (T22) given based on the probability that the surname appears will be listed together. This index can be interpreted as a “index of probability of separation” based on statistics.

  Similarly, FIG. 5B is an example showing the configuration of the first name surname history 162. In this example, a list (T23) of the same surname as the surname / name dictionary 161 and an index (T24) whose surname is a numerical value actually appearing in this process are listed together. This index can be interpreted as an “index of probability of separation” based on actual use. As will be described later, the first and last name dictionary 161 may be arbitrarily replaced, and the first and last name history 162 may be updated as appropriate by taking in and adding the history of other devices.

  Next, the processing flow of this embodiment will be described in more detail with reference to FIG. First, the personal authentication medium 200 possessed by the user to be input is photographed by the camera 121 built in the information input device 100 by the photographing means 120 of the information input device 100. The photographing unit 120 stores photographed image data.

  Subsequently, the image dividing unit 130 of the information input device 100 acquires information on which layout information 131 record is used (which type of personal authentication medium layout information is used). The operator of the information input apparatus 100 may instruct layout information to the control means 110 by manual input via the input unit 104, and characters are input from each record of the held layout information 131. Is extracted from the captured image data, the image data relating to the region is obtained, the density analysis and the spatial frequency analysis are performed, and the accuracy of the presence of characters from the image data is confirmed. The record of the highest layout information 131 may be automatically selected and set. The image dividing unit 130 divides the image data into a plurality of partial image data using predetermined layout information.

  The image correction unit 140 acquires partial image data related to the first and last names and partial image data related to the address among the partial image data divided by the image dividing unit 130, and first calculates the number of characters of the first and last names using a spatial frequency analysis or the like. Thus, from the number of characters, for example, the character width is determined to be full-width if it is 7 characters or less, 2 / 3-angle if it is 14 characters or less, and half-width if it is 15 characters or more. Subsequently, partial image data relating to the address is obtained, the width for one character is obtained, the character width is determined by comparison with the character width of the first and last names, and the width of one character is full-width (predetermined size) Each partial image data is expanded in the width direction (detailed processing flow will be described later).

  FIG. 7 is a diagram showing an example of the appearance of a personal authentication medium 200 such as a public identification card. In this example, since the number of characters detected from the partial image data relating to the first and last names is four, the image correcting unit 140 determines that the character width 201 of the partial image data relating to the first and last names is full-width. Further, the character width 202 detected from the partial image data related to the address is determined to be 2/3 corners here, as compared with the character width 201 detected from the partial image data related to the surname. Therefore, the partial image data related to the first and last names is not corrected as it is, and the partial image data related to the address is enlarged in the width direction so that the width of one character becomes full-width.

  FIG. 8 is a flowchart showing a detailed processing flow related to the correction of the character size in the image correcting unit 140. First, the image correction unit 140 acquires partial image data relating to the first and last names (S201). Subsequently, the number of characters of the first and last names is acquired using spatial frequency analysis or the like (S202). Next, as described above, it is determined from the acquired number of characters whether the character width is full-width, 2 / 3-angle, or half-width (S203, S204 to S206). Subsequently, partial image data relating to the address is acquired (S207), and the character width of the address is acquired using spatial frequency analysis or the like (S208). Here, it is determined whether the character width of the address is full-width, 2 / 3-square, or half-width compared with the character width of the first and last name determined previously (S209, S210 to S212). Based on the character width obtained here, the size of the partial image data related to the first and last names is corrected so that the character width becomes full-width (S213), and the size of the partial image data related to the address is changed to full-width (S214), and the process is finished so that the character widths in the partial image data relating to the first name and the last address are the same.

  Returning to FIG. 6, the character recognition unit 150 of the information input device 100 converts each corrected partial image data into text data using optical character recognition (OCR). Here, since optical character recognition (OCR) processing is widely known and commercially available, the apparatus according to the present embodiment is mounted with an existing OCR. At this time, since it is desired to use a general-purpose OCR, the processing described in the above-described image correction unit 140 is performed, and normalization is performed so that the character width is unified to full width, so that the input data is generalized.

  Subsequently, the first and last name classification means 160 classifies the first and last names converted into text data by the character recognition means 150 into first and last names using the first and last name dictionary 161 and the first and last name history 162.

FIG. 9 is a flowchart for obtaining an index for separating first and last names in the first and last name classification means 160. First, a character string of first and last names converted into text data is acquired (S301). Next, the number of positions that are likely to be delimiters of the first and last names is obtained from the acquired first and last character strings (S302). For example, if there is a user whose first and last names are “Riko Hatano”, positions that may be delimited Are the four places indicated by the following "/".
Wave / Ta / No / Ri / Koko.
Here, this number “4” is looped (S303), and four surname patterns are created (S304). There are four types of surname patterns to be created: “Hami”, “Hata”, “Hatano”, and “Ri Hatano”. Here, the number of surname patterns created is looped (S305), and the surname / name dictionary is inquired as to whether the surname pattern exists in the surname / name dictionary 161 (S306, S307). If the surname exists, the index (T22) corresponding to the surname field (T21) of the surname dictionary 161 is acquired and held as an index value of the probability of separation (S308). Subsequently, the surname surname dictionary is inquired as to whether the surname exists in the surname history 162 (S309, S310). If the surname exists, the index (T24) corresponding to the surname field (T23) of the surname history 162 is acquired and added to the index value of the probability of the previous separation (S311). When the surname record 161 and the surname record held in the surname name history 162 are synchronized (always the same), the process of querying the surname / name dictionary (S309, S310) is unnecessary, and the index of the surname / name dictionary is retained (S308). ) Immediately after that, a surname history index may be added (S311). In this way, the index of the certainty of how to separate each surname pattern is stored (S312), and the process ends.

  Returning to FIG. 6, the classification determination unit 170 of the information input device 100 modifies the character string of the first and last names converted into text data based on the index of the probability of separation for the last name pattern thus created. Displayed on the display unit 105 of the apparatus 100, and the input indicating the separation between the first name and the last name by the user is detected from the input unit 104 of the information input apparatus 100, and the text data is classified into the text data of the last name and the text data of the first name. To do. In this example, since a tablet PC equipped with a touch panel display is assumed, the input unit 104 and the display unit 105 are a single superimposed device.

  FIG. 10 is an example of first and last name texts displayed on the display unit 105 of the information input device 100 by the classification determining unit 170 in the present embodiment. FIG. 10A shows an example in which the first and last name text is displayed as it is regardless of the index of the probability of separation. The user determines the separation between the first name and the last name by, for example, separating the first name and the last name by using a gesture of cutting from top to bottom with a finger. The gesture of cutting from the top to the bottom with a finger may be detected by the touch panel that is the input unit 104 of the information input device 100. FIG. 10B is an example in which the first and last name texts are displayed by adjusting (modifying) the interval between the first name and the last name based on an index of the probability of separation. In this example, the space between characters with high probability of separation is displayed widely (Riko Hatano is more likely than Riko Hatano), and gesture operations are easier. Furthermore, as shown in FIG. 10C and FIG. 10D, based on an index of the probability of separation, a modification such as a symbol or a mark is given between the last name and the first name, making it easy to perform operations by gestures. You may make it do.

  Returning to FIG. 6, in addition to the text of the surname and first name in which the classification is determined in this way, the classification determination unit 170 is divided into partial image data based on layout information such as an address, date of birth, validity period, and the like, and has been recognized. Various text data may be displayed on the display unit 105 of the information input device 100.

  Next, the data storage unit 180 of the information input device 100 is divided into partial image data according to layout information such as the above-described divided first and last name text, address, date of birth, and validity period, and character recognition is performed. Various text data are stored and stored as a personal information DB (database) 181.

  The information input device 100 can also transmit information recorded in the personal information DB 181 to the member server 300 of an external company via the communication unit 106 and the network 500. .

  According to the present embodiment, personal information described in a personal authentication medium possessed by a user is photographed, and the image data is accurately extracted by optical character recognition (OCR). There is provided an information input device capable of accurately distinguishing between first name and last name with an easy operation even when there is no clear separation (blank) between first names. This device is portable and can be used in small event venues, regardless of installation location, and it can also be used for age verification by calculating the current age from the date of birth recognized from the personal authentication medium. Can be completed on the spot.

  The present invention is not limited to the above-described embodiments, and various modifications and applications are possible within the scope of the claims. For example, the information input device 100 may be a device such as a normal PC instead of the tablet PC as shown in the present embodiment. In this case, the present invention can be implemented in the same manner even if the mouse is used to input an instruction such as a gesture for cutting from the top to the bottom with the finger in the classification determining unit 170.

DESCRIPTION OF SYMBOLS 100 Information input device 110 Control means 120 Image | photographing means 130 Image division means 131 Layout information 140 Image correction means 150 Character recognition means 160 First name surname classification means 161 First name surname dictionary 162 First name surname history 170 Classification determination means 180 Data storage means 181 Personal information DB
200 Medium for personal authentication 300 Corporate member server 500 Network

Claims (4)

Photographing means for photographing a public identification card and storing it as image data;
Image dividing means for dividing the stored image data into a plurality of partial image data using layout information determined for each type of the public identification card;
Image correction means for enlarging or reducing the partial image data to a predetermined size;
Character recognition means for storing each of the partial image data scaled up to the predetermined size into text data using optical character recognition and holding it together with attributes obtained from layout information;
First and last name classifying means for classifying a character string obtained by converting partial image data relating to first and last names into text data;
A classification determining means for detecting an input indicating a separation between a first name and a last name of a first and last name and determining a text data as a separator between the last name text data and the first name text data;
Data storage means for storing each text data,
The image correcting means determines the character width by comparing the number of characters of the first and last names included in the partial image data relating to the first and last names and the character width of one character in the partial image data relating to the address, and thereby determines the partial image data. An information input device characterized by enlarging or reducing the size so as to have the same character width. Photographing means for photographing a public identification card and storing it as image data;
Image dividing means for dividing the stored image data into a plurality of partial image data using layout information determined for each type of the public identification card;
Image correction means for enlarging or reducing the partial image data to a predetermined size;
Character recognition means for storing each of the partial image data scaled up to the predetermined size into text data using optical character recognition and holding it together with attributes obtained from layout information;
First and last name classifying means for classifying a character string obtained by converting partial image data relating to first and last names into text data;
A classification determining means for detecting an input indicating a separation between a first name and a last name of a first and last name and determining a text data as a separator between the last name text data and the first name text data;
Data storage means for storing each text data,
The first and last name classifying means classifies the character string obtained by converting the partial image data relating to the first name to the text into one or more first names and first names, and obtains an index of the probability of each division,
The classification determining means modifies the character string based on these indices and displays it on a display unit. The image correcting means determines the character width by comparing the number of characters of the first and last names included in the partial image data relating to the first and last names and the character width of one character in the partial image data relating to the address, and thereby determines the partial image data. The information input device according to claim 2, wherein the information is enlarged / reduced to have the same character width. A computer program for causing a computer to function as the information input device according to any one of claims 1 to 3.

Images