JP2008040910A - Character string determining device, character string determining method, character string determination program and computer readable recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a character string determining device determining a predefined type to which a target character string corresponds with less throughput. <P>SOLUTION: The character string determining device is provided with a character string pattern detecting part 22a for detecting whether a character string pattern associated with a character string item and showing a specific shape is included in character string information, and an item determining part 22b for defining an item of a character string corresponding to a character string pattern as an item of a corresponding target character string when the character string pattern detecting part 22a detects the character string pattern in the character string information. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a character string determination device that determines the type of a target character string.

  Currently, optical character reader (OCR) technology is realized and used as various character readers such as document readers and form readers. Among them, many devices, such as a business card reader, determine what each character string item such as a name, an address, and a company name written on a business card is.

  On the other hand, the OCR has been downsized in recent years, and a camera-equipped mobile phone that takes a picture of an object to be read by the camera and recognizes it has been released. The OCR mounted on these mobile phones generally has a specification that the recognition target is recognized after each line is photographed, and the recognition target is an email address, URL (uniform resource locator), telephone number, etc. Some of them are alphanumeric characters only, while others recognize Japanese character strings such as names and addresses.

  The main use of these OCRs mounted on mobile phones is to register names, phone numbers, e-mail addresses, URLs, etc. described on business cards or other documents in the phone book of the mobile phone. is there.

  The OCR mounted on the camera-equipped mobile phone generally shoots the entire business card at once, and does not recognize all the items of name, address, and company name written on the business card. One of the reasons for the specification of recognition is that there is a problem with the processing capability of mobile phones. The problem will be described below.

  Patent Document 1 describes a difference in processing speed between a PC (personal computer) and a mobile phone. According to this, when recognizing an image (one image) with one line and 33 characters written as "http://www.kaisya.co.jp/index.html" In contrast to the processing time, a personal computer (with Pentium (registered trademark) III) requires a processing time of 0.1 seconds.

  For example, in the case of a specification in which the shooting range of the camera is expanded and, for example, one business card is photographed and all items described on one business card are recognized, 165 characters are written in one business card. From the above processing speed value, it can be said that it takes 8 seconds (1.6 ÷ 33 × 165) processing time by simple calculation.

  The above calculation is an estimated value when the recognition target is English, and when a business card written in Japanese is targeted, a Japanese character string such as name and address must be recognized. Therefore, in general, it is estimated that the recognition processing time further increases.

  As described above, since a small information terminal such as a mobile phone has a low processing capability, there is a problem that a processing time is long even when a document having a size of about one business card is recognized. It can be said that this is a problem to be solved when installing the.

  As one method for solving the problem that the processing time is long, a method of limiting the recognition target to the items desired by the user instead of all items of the business card is conceivable.

  One of the main uses of OCR for camera-equipped personal digital assistants such as mobile phones is the phone book etc. that the personal digital assistant has the read information (name, address, company name, telephone, email address, etc.) However, if the OCR is installed in a mobile phone or the like, the recognition function limited to name, phone, and e-mail address is considered to be sufficiently useful.

  Therefore, it can be said that the method of limiting the recognition target items to the necessary items, automatically extracting only the necessary items from the recognition target, and recognizing them is an effective solution.

  Hereinafter, as a specific example of the above solution, a method for extracting items necessary for telephone directory registration from a business card will be described. The following methods can be considered as methods for automatically extracting a name, a phone number, and an e-mail address used for registering a phone book from a business card.

  In order to take out the first and last name line from the business card, for example, as described in Patent Document 2, it is only necessary to take out a character string laid out in the center of the business card, and it can be extracted relatively easily.

  On the other hand, it is difficult to extract a telephone line and a mail address line from a business card with only layout information. Therefore, for example, a determination and extraction method can be considered in which it is determined whether a character string is an alphanumeric character string or a Japanese character string, and if the character string is an alphanumeric character string, it is a telephone line or a mail line.

An example of a technique for determining whether a character string is an alphanumeric character string or a Japanese character string is described in Patent Document 3. In the character recognition method described in Patent Document 3, the complexity of characters in a character string is measured to determine whether it is an alphanumeric character string or a Japanese character string. Using this method, after first extracting the character string described on the business card, it is determined whether the character string is a Japanese line or an English line. It is conceivable to perform a process of recognizing the column in more detail and not recognizing the Japanese line.

As an example of a technique for extracting a rectangle circumscribing an area where dots constituting characters are connected and extracting a character string using the extracted rectangle, a technique described in Patent Document 4 can be cited.
JP 2005-182772 A (published July 7, 2005) JP-A-6-103402 (published on April 15, 1994) JP 62-22186 (published January 30, 1987) Japanese Patent Laid-Open No. 5-81474 (published April 2, 1993)

  However, the character recognition method described in Patent Document 3 determines whether a recognition target is a general Japanese character string or an English character string. Whether it is a character string of a type, for example, a telephone number or an e-mail address cannot be uniquely determined. In addition, it is impossible to narrow down which of the predetermined character string items (types) the target character string corresponds to.

  Therefore, in order to determine the type of the target character string, there arises a problem that recognition processing such as pattern matching, which requires a large amount of processing, must be performed.

  The present invention has been made to solve the above-mentioned problems, and its purpose is to determine which of the predetermined types the character string to be recognized corresponds to with a small amount of processing. An object of the present invention is to provide a character string determination device capable of

  In order to solve the above-described problem, the character string determination device according to the present invention acquires character string information indicating the shape of a character constituting the target character string, and based on the character shape indicated by the character string information. Whether or not the character string information includes a character string pattern that is information indicating a specific shape associated with the character string type. When the character string pattern is detected in the character string information by the specific character detection means to detect and the specific character detection means, the type of the character string corresponding to the character string pattern is It is characterized by comprising a type determining means for making a type.

  In order to solve the above-described problem, the character string determination method according to the present invention acquires character string information indicating the shape of a character constituting the target character string, and based on the character shape indicated by the character string information. Is a character string determination method in a character string determination device that determines the type of a target character string, and the character string information includes a character string pattern that is associated with the character string type and indicates a specific shape When the character string pattern is detected in the character string information by the specific character detection step for detecting whether or not the character string is detected, the type of character string corresponding to the character string pattern is And a type determination step for setting the type of the target character string.

  According to said structure, a specific character detection means detects whether a character string pattern is contained in character string information, and when it is contained, a kind determination means respond | corresponds to the said character string pattern It is determined that the type of the character string is the type of the target character string.

  Therefore, it is possible to determine the type of the target character string by detecting whether or not the character string information includes the character string pattern, and recognize which character each character of the target character string is. Compared to the configuration, it is possible to reduce the processing amount for determining the type of the target character string.

  Therefore, even when incorporated in a small portable terminal having a low processing capability such as a mobile phone, it is possible to realize a character string determination device that can quickly determine a target character string.

  Note that the above characters include symbols used when writing a document, such as parentheses, commas, colons, and semicolons.

  The character string pattern preferably includes information on a character having a specific shape, information on a character including a component having a specific relative positional relationship, or both.

  According to the above configuration, since the character string pattern includes information on a character having a specific shape and / or information on a character including a component having a specific relative positional relationship, the character string pattern is detected. The amount of processing required to do this is smaller than the amount of processing in the method of identifying each character by matching using a dictionary.

  Therefore, the processing amount for determining the type of the target character string can be further reduced.

  Further, the character string determination device refers to the character string information by referring to a plurality of character recognition dictionaries respectively corresponding to the type candidates when the type determination unit lists a plurality of types of candidates. It is preferable to further include character recognition means for identifying included characters.

  According to the above configuration, when a plurality of candidate types of the target character string are listed by the type determining unit, the character recognizing unit refers to the plurality of character recognition dictionaries respectively corresponding to the plurality of candidates. Identifies characters contained in column information. In other words, in the above configuration, the type of the target character string is narrowed down by the type determination unit, and the character recognition unit identifies the character with respect to the narrowed down type.

  Therefore, the character recognition dictionary having a small capacity can be used by the character recognition means, the processing amount for character recognition can be reduced, and the processing speed can be increased.

  Further, the character string determination device, when a plurality of the types of candidates are listed by the type determination means, and the target character string of the type that matches the specified type that is the type of the specified target character string, It is preferable to further include character recognition means for identifying a character included in the character string information by referring to a character recognition dictionary corresponding to the specified type when included in a plurality of candidates.

  According to the above configuration, the character recognizing unit may include a target character string of a type that matches the specified type when a plurality of candidates for the type of the target character string are listed by the type determining unit. If it is included, the character included in the character string information is identified by referring to the character recognition dictionary corresponding to the designated type. That is, the type of the target character string is narrowed down by referring to the designated type information, and the character recognition means identifies the character with respect to the narrowed down type.

  Therefore, the character recognition dictionary having a small capacity can be used by the character recognition means, the processing amount for character recognition can be further reduced, and the processing speed can be further increased.

  Moreover, it is preferable that the said character string determination apparatus is further provided with the character string specific | specification means which specifies the kind of the said object character string based on the character recognition result of the said character recognition means.

  According to the above configuration, the character string specifying unit specifies the type of the target character string by using the character recognition result of the character recognition unit. Since the character recognition result of the character recognition means indicates what character is included in the character string information, the type of the target character string can be specified as one.

  The technical scope of the present invention also includes a character string determination program for causing a computer to function as each means of the character string determination device and a computer-readable recording medium on which the character string determination program is recorded.

  As described above, the character string determination device according to the present invention is a specification that detects whether or not a character string pattern that is information indicating a specific shape associated with a character string type is included in the character string information. When the character string pattern is detected in the character string information by the character detection means and the specific character detection means, the type of the character string corresponding to the character string pattern is set as the type of the target character string. And a type determination unit.

  As described above, the character string determination method according to the present invention is a method for detecting whether or not a character string pattern, which is information indicating a specific shape and associated with a character string type, is included in the character string information. When the character string pattern is detected in the character string information by the character detection step and the specific character detection means, the type of the character string corresponding to the character string pattern is set as the type of the target character string. And a type determination step.

  Therefore, there is an effect that the processing amount for determining the type of the target character string can be reduced.

  One embodiment of the present invention will be described below with reference to FIGS. Hereinafter, the mobile phone 1 will be described as an example of the character string determination device of the present invention, but the character string determination device of the present invention is not limited to a mobile phone. The cellular phone 1 captures an image of a recognition target document such as a business card, acquires information on a plurality or a single character string (character string information) included in the captured image, and shapes the characters indicated by the character string information. The type of the target character string is determined based on the above. The mobile phone 1 has a function of registering information such as a telephone number and an e-mail address included in the target character string in a telephone directory that is a database.

(Configuration of mobile phone 1)
FIG. 2 is a functional block diagram showing the configuration of the mobile phone 1 (character string recognition device) of the present embodiment. As shown in FIG. 2, the mobile phone 1 includes a main control unit 2, a temporary storage unit 3, a storage unit 4, an input control unit 5, an operation unit 6, an output control unit 7, a display unit 8, and a camera 9. . These units are connected by a data bus.

  The mobile phone 1 includes members necessary for functioning as a mobile phone, such as a communication unit, a microphone, and a speaker, but these members are omitted in FIG.

  The main control unit 2 controls the mobile phone 1 and includes a character string determination unit 2a and a character string determination control unit 2b.

  The character string determination unit 2a includes a character string extraction unit 21, a specific character determination unit 22, a character string recognition unit 23 (character recognition unit), and an item determination unit 24 (character string identification unit). Details of these parts will be described later.

  The character string determination control unit 2b controls each part of the character string determination unit 2a.

  The temporary storage unit 3 includes a semiconductor memory such as a RAM (Random Access Memory), and has a function of temporarily storing data. The temporary storage unit 3 includes a target document storage unit 31, a target character string storage unit 32, a target item storage unit 33, and a recognition result storage unit 34.

  The target document storage unit 31 temporarily stores a document image to be recognized (hereinafter referred to as a target image).

  The target character string storage unit 32 stores character string information (character string information) (hereinafter referred to as a target character string) extracted by the character string extraction unit 21 from the target image stored in the target document storage unit 31. Memorize temporarily.

  The target item storage unit 33 stores a character string item (hereinafter referred to as a recognition target item) designated by the user.

  The recognition result storage unit 34 temporarily stores the recognition result of the character string recognition unit 23.

  The storage unit 4 includes a ROM (Read Only Memory), a hard disk, or the like, and has a function of (permanently) storing data. The storage unit 4 includes a character determination rule table 41, a corresponding item table 42, a character recognition dictionary 43, and a telephone book 44.

  The character determination rule table 41 is a rule for determining whether or not a target character string stored in the target character string storage unit 32 includes a specific character string pattern (hereinafter simply referred to as a character string pattern). This is a table to be defined and used by the specific character determination unit 22. Details of the character determination rule table 41 will be described later. Further, as will be described later, the character string pattern is information indicating a specific shape associated with the type of character string, and has information on characters having a specific shape or a specific relative positional relationship. It includes information on characters including, or both.

  The corresponding item table 42 shows the correspondence between the character string pattern included in the target character string and the item of the target character string. That is, the character string pattern is associated with the type of character string by the corresponding item table 42. The specific character determination unit 22 refers to the correspondence item table 42 and detects a character string pattern included in the target character string. Details of the corresponding item table 42 will be described later.

  The character recognition dictionary 43 is a character pattern matching dictionary used by the character string recognition unit 23.

  The telephone book 44 is a database used by the user, and stores information such as name, telephone number, FAX number, mail address, URL, and address registered by the user. The writing to the telephone directory 44 is performed by the main control unit 2.

  The input control unit 5 detects an input operation of the operation unit 6 by the user, and transmits a command such as selection, instruction, and determination to the main control unit 2. In particular, the input control unit 5 receives information on a recognition target item (specified type) that is an item of a target character string specified by the user. The input control unit 5 includes a target item instruction unit 51 (accepting means).

  The target item instruction unit 51 receives an instruction of the recognition target item input from the user through the operation unit 6 and transmits the instruction information to the main control unit 2.

  The operation unit 6 is an input device for receiving a user instruction and outputting the instruction to the input control unit 5. For example, there are various types of character recognition devices such as input keys. For example, a key constituting a so-called keyboard, a cross key, a numeric key, and the like can be used as the operation unit 6.

  The output control unit 7 receives various information output from the main control unit 2, for example, a recognition result of the target character string, and outputs the information via the display unit 8.

  The display unit 8 is a display device of an image display system such as a liquid crystal display (LCD (Liquid Crystal Display)), and can display information output from the output control unit 7.

  The camera 9 is a CCD (Charge Coupled) camera, for example, and images a character string to be recognized. The target image captured by the camera 9 is stored in the target document storage unit 31 via the input control unit 5 and the main control unit 2.

(Details of each part of the character string determination unit 2a)
Next, functional blocks provided in the character string determination unit 2a will be described in detail.

  The character string extraction unit 21 extracts a target character string from the target image stored in the target document storage unit 31, and outputs the extracted target character string to the target character string storage unit 32. The output target character string is temporarily stored in the target character string storage unit 32. The method for extracting the character string is not particularly limited, and any conventionally used method may be used. In the present embodiment, a method of extracting a rectangle (a connected character block) and extracting a character string using the extracted rectangle is used. An example of such an extraction method is described in Patent Document 4. The character string information output from the character string extraction unit 21 will be described later.

  FIG. 1 is a functional block diagram illustrating a configuration of the specific character determination unit 22.

  The specific character determination unit 22 determines the target character string stored in the target character string storage unit 32, that is, the item of the target character string extracted by the character string extraction unit 21.

  As shown in FIG. 1, the specific character determination unit 22 includes a character string pattern detection unit 22a (specific character detection unit), an item determination unit 22b (type determination unit), and an item comparison unit 22c (item comparison unit). Yes.

  The character string pattern detection unit 22 a detects a character string pattern included in the target character string stored in the target character string storage unit 32. Specifically, the character string pattern detection unit 22a refers to the corresponding item table 42 and sequentially detects whether or not a character string pattern associated with a character string item is included in the target character string. A method for determining whether or not the target character string includes a character string pattern is described in the character determination rule table 41 of the storage unit 4, and the character string pattern detection unit 22 a uses the character determination rule table 41. The character string pattern is detected while referring to it.

  The item determination unit 22b refers to the corresponding item table 42 and determines an item corresponding to the character string pattern detected by the character string pattern detection unit 22a as an item of the target character string. In other words, when the character string pattern is detected in the target character string by the character string pattern detection unit 22a, the item determination unit 22b selects the item of the character string corresponding to the character string pattern as the target character string. It is an item.

  The item comparison unit 22 c compares the item determined by the item determination unit 22 b with the recognition target item stored in the target item storage unit 33.

  The character string recognition unit 23 refers to the character recognition dictionary 43 for characters included in the target character string stored in the target character string storage unit 32, that is, the target character string extracted by the character string extraction unit 21. recognize. At this time, the character string recognition unit 23 is included in the target character string by referring to the character recognition dictionary corresponding to the item candidate determined by the item determination unit 22b or the character recognition dictionary corresponding to the recognition target item. Identify characters.

  The character string recognizing unit 23 recognizes the recognition target item received by the input control unit 5 particularly when the item determining unit 22b does not determine one item of the target character string (when a plurality of item candidates are listed). If the target character string of the item that matches is included in the plurality of candidates, recognition processing is performed. Note that the character string recognition unit 23 performs a recognition process to recognize the content (specifically, the character code) of the target character string even when there is one candidate for the above item.

  The method of character identification by the character string recognition unit 23 is not particularly limited, and any conventionally used method may be used. The character string recognition unit 23 outputs a recognition result of characters included in the target character string to the recognition result storage unit 34. The output recognition result is stored in the recognition result storage unit 34.

  The item determination unit 24 uses the recognition result of the character included in the target character string stored in the recognition result storage unit 34, that is, the recognition result of the character output by the character string recognition unit 23 to use the target character string. One item is specified. The item determination method in the item determination unit 24 is not particularly limited, and any conventionally used method may be used. The item determination unit 24 outputs the determination result to the character string determination control unit 2b.

(Documents to be recognized by the mobile phone 1)
Next, the recognition target document of the mobile phone 1 will be described with reference to FIG. FIG. 3 is a diagram illustrating an example of a recognition target document of the mobile phone 1.

  A business card can be mentioned as an example of the recognition target document of the mobile phone 1. As shown in FIG. 3, a business card given as an example has “△ yama × m” as the name, “General Affairs Department” as the department to which it belongs, and “XYZ Corporation” as the company name. Further, a telephone number, an e-mail address, and a home page URL are described in the lower part of the business card.

  The recognition target document of the mobile phone 1 is not limited to a business card, and may be a document such as a letter or an advertisement.

(Examples of items displayed on the display unit 8)
Next, an example of the item of the target character string displayed on the display unit 8 will be described with reference to FIG. FIG. 4 is a diagram illustrating an example of items of the target character string displayed on the display unit 8.

  As shown in FIG. 4, the mobile phone 1 displays a plurality of items (1 to 5 in FIG. 4) on the display unit 8, and waits for an instruction from the user regarding which item character string to recognize.

  When the user presses any one of the numeric keys 1 to 5 on the operation unit 6, a set of any one of the target items 1 to 5 is determined. For example, when the user desires to recognize a character string limited to items such as name, telephone, and mail, the combination of these items (recognition target item) is indicated by the number 4, and therefore the numeric key 4 is pressed. That's fine.

  The combination of a some item is not limited to what was shown in FIG. 4, What kind of combination may be sufficient.

(Details of the corresponding item table 42)
Next, the correspondence item table 42 will be described with reference to FIGS. 5 and 6. The corresponding item table 42 shows the correspondence between the character string pattern included in the target character string and the item of the target character string. 5 and 6 are diagrams showing an example of the correspondence item table 42. FIG.

  As shown in FIGS. 5 and 6, in the correspondence item table 42, the character string pattern and the corresponding item of the target character string are written on the same line, and the target character string includes a certain character string pattern. If so, it indicates that the target character string corresponds to an item (part of an item group) corresponding to the character string pattern. Each character string pattern is given a serial number (1 to 6 in FIG. 5). 5 and 6, the column surrounded by the broken line on the right side of the corresponding item table 42 shows an example of the target character string, and is not information in the corresponding item table 42.

  The correspondence item table 42 can be roughly divided into two types according to the contents of the character string pattern, as distinguished from FIGS. 5 and 6. That is, one has one item corresponding to the included character string pattern (corresponding item table 42a in FIG. 5), and the other has two or more items corresponding to the character string pattern (FIG. 5). 6 corresponding item table 42b).

  The corresponding item table 42a is a table in which the item of the target character string is determined when a character string pattern exists in the target character string.

  For example, as shown in FIG. 5, when the target character string includes character string patterns of “L”, “:” and parentheses ((), the item of the target character string is “phone”. .

  The corresponding item table 42b is a table used when the corresponding item table 42a cannot be applied, that is, when the item of the target character string cannot be specified as one.

  For example, as shown in FIG. 6, when the target character string includes character string patterns of “L” and “:”, the items of the target character string are “phone”, “mail”, “URL” It will be either.

  As described above, when performing item determination, first, it is checked whether or not one item can be specified using the corresponding item table 42a. If one item cannot be specified, the item is determined using the corresponding item table 42b. You should narrow down.

  In addition, the character string pattern used for item determination is not limited to what was shown in FIG.5 and FIG.6.

(Details of character determination rule table 41)
Next, the character determination rule table 41 will be described with reference to FIG. FIG. 7 is a diagram illustrating an example of the character determination rule table 41. The character determination rule table 41 defines rules for determining whether or not a character string pattern is included in the target character string. The character string pattern detection unit 22a detects a character string pattern according to the rules described below.

  In FIG. 7, each rule is expressed in a sentence so that it can be easily understood. However, each rule is actually expressed in a form that can be recognized by a computer.

  As shown in FIG. 7, the character determination rule table 41 indicates what character (including the number of constituent rectangles) or character relative position in the character string is recognized as a character. It shows a judgment rule. That is, the character string pattern includes a character having a specific shape or a character including a component having a specific relative positional relationship. The character determination rule table 41 includes a character shape or positional relationship, a character determination result, Are associated with each other.

  Hereinafter, the determination method for each character will be specifically described with reference to FIGS. FIG. 8 is a diagram illustrating the shape of a colon. FIG. 9 is a diagram showing the shape of parentheses. FIG. 10 is a diagram showing the shape of the slash.

  As shown in FIG. 8, the colon is composed of two rectangles (components), and both the two rectangles are small rectangles. Actually, the determination may be made on the basis that the width and height of the two rectangles constituting the character are less than ((character string height) × (threshold 1)).

  The colon character is elongated. Actually, the determination is made using a criterion of character height> character width × (threshold value 2). In addition, the two rectangles need to be substantially overlapped by scanning with the vertical scanning line.

  If this is explained using the example shown in FIG. 8, it is only necessary to check whether α1 is not so small as compared to α2. Actually, the determination is made using a threshold value. Α1 is the distance from the more right end of the left upper end of the colon and the left end of the lower step to the more left end of the right end of the upper step and the right end of the lower step, α2 Is the distance from the left end of the upper step portion of the colon and the left end of the lower step portion to the more right end of the right end of the upper step portion and the right end of the lower step portion.

  Although the determination using a threshold value or the like is actually applied to the characters described below, only the outline of the determination rule will be described.

  As shown in Fig. 9, regarding the left parenthesis ((), it is only necessary to check whether there is a character component in the upper left corner, lower left corner, and right middle (the dashed square portion in Fig. 9). The same applies to)).

The slash sequence “//” is determined by running a scanning line from left to right. As shown in FIG. 10, the distance from the left end of the character to the first character component is β1, the distance from the first character component to the second character component is β2, the distance from the second character component to the right end of the character Is β3,
As β1 moves from the upper line to the lower line, it decreases almost monotonously.
β3 increases almost monotonically as it moves from the upper line to the lower line.
β2 is substantially constant regardless of which line the scan line is on.
If it has the above characteristics, it will be determined as a character consisting of a series of slashes.

  The slash can be determined by the state of increase or decrease in the distance from the left end to the character component and the distance from the character component to the right end in the scanning line from the left to the right as in the case of the slash sequence “//”.

  “L” has a large peak at the left end when the character component histogram is taken in the horizontal direction, and is almost constant thereafter. In addition, when the character component histogram is taken in the vertical direction, there is a large mountain at the base portion which is almost constant except for the base portion. By taking such a histogram, “L” is determined.

  The period may be determined using a criterion that one small rectangle exists below the character string.

  The hyphen may be determined using the criterion that a single horizontally long rectangle exists in the middle of the character string.

  The number, l (el), consists of a single rectangle, and can be easily determined by criteria such as the vertical width being larger than the horizontal width.

  Actually, “l (el)” and “1 (壱)” are further vertically longer than “0”, “2”, “3”, etc., and therefore it is preferable to set individual determination thresholds. .

  Also, it is difficult to determine “story”, “tongue”, etc. by shape and position. Therefore, character recognition is performed using a recognition dictionary of only “story” and “tongue”, and when the reliability is 0.9 or more (maximum is 1), it can also be determined as “talk” or “tongue”. .

  The determination rule for each character is not limited to the above, and the determination target character itself is not limited to the above.

Note that there is a possibility that the characters of “story” are separated into two characters of “word” and “tongue”.
As a countermeasure, a rule of “tongue” + “:” may be added.

(Structure of character string information extracted by the character string extraction unit 21)
Next, an example of character string information extracted by the character string extraction unit 21 will be described with reference to FIG. FIG. 11 is a diagram illustrating an example of the structure of the character string information 60 extracted by the character string extraction unit 21. FIG. 11 shows character string information for one character string.

The character string information includes character number information 61 indicating how many characters are included in the character string.
It consists of character string coordinate information 62 indicating the upper left and lower right coordinates of the character string, and character information 63 regarding each character included in the character string.

  Further, the character information 63 includes character rectangle number information 64 which is the number of rectangles constituting the character, character coordinate information 65 indicating the upper left and lower right coordinates of the character, and character rectangle information 66 for each character. The character rectangle information 66 includes rectangle coordinates corresponding to the rectangles constituting the character.

  Although it has been described above that the conventional technique may be used for extracting the character string, since there are separated characters and the like in Japanese, 1 is used without using the reliability (similarity or the like) of characters as a result of character pattern matching. It is difficult to determine characters correctly. Therefore, in this embodiment, the character rectangle group which overlaps by the scanning by a vertical direction scanning line is made into one character.

This extraction method will be described using an example of extracting an e-mail address written on a business card or the like. here,
Email: sharp.taro@sharp.co.jp
Recognize the first 5 characters.

  Of the characters included in this character string, “me”, “−”, “s” are recognized as one rectangle per character. “Le” is recognized as two rectangles, “No” on the left side and “Le” on the right side, but these rectangles are treated as different characters because they do not overlap when scanned by the vertical scanning line.

  “:” Is composed of two points, an upper point and a lower point, but is overlapped by scanning by a vertical scanning line, so that it is recognized as one character by two rectangles of the upper point and the lower point.

(Character recognition method using the character recognition dictionary 43)
Next, an example of how to use the character recognition dictionary 43 by the character string recognition unit 23 will be described.

For example, when it is determined that the item of the target character string is “phone”, in addition to numbers 0 to 9, hyphens, and parentheses, key words (telephone, TEL, Tel: etc.)
"Den", "Talk", "T", "E", "L", "t", "e", "l", ":"
Only characters such as can be used for dictionary matching.

Similarly, when it is determined that the item of the target string is “mail”, in addition to alphanumeric characters, @, period, hyphen, and underscore, characters in commonly used key words (email: etc.), that is, ,
"Email",":",
Only characters such as can be used for dictionary matching.

  When the target character string item is specified as one item, it is not necessary to specify the item, but since the content of the target character string (specifically, the character code) is unknown, the character recognition dictionary is usually used. It is necessary to determine the contents of the target character string (specifically, the character string is recognized).

  At this time, it is possible to speed up matching by using only necessary dictionary categories.

  If the target character string is not uniquely determined, the sum of characters to be matched corresponding to the item may be used. For example, if the item is either telephone or FAX and is not uniquely determined, both the dictionary character category to be checked when the item is a phone call and the dictionary character category to be checked when the item is FAX may be used. .

(Processing flow in the mobile phone 1)
Next, the flow of processing for extracting a plurality of character strings from the target image and determining the item of each character string in the mobile phone 1 will be described with reference to FIG. FIG. 12 is a flowchart showing the flow of processing in the mobile phone 1. In the following, the flow of processing when the specific character determination unit 22 cannot determine one item of the target character string will be described.

  First, when a user inputs an instruction for a recognition target item via the operation unit 6, the instruction is transmitted to the target item instruction unit 51 (S1). And the target item instruction | indication part 51 outputs the said instruction | indication (information of recognition target item) to the main control part 2 (reception process).

  When receiving the above instruction, the main control unit 2 stores information on the recognition target item in the target item storage unit 34.

  Further, the main control unit 2 captures a recognition target document by controlling the camera 9, and stores the target image, which is the captured image, in the target document storage unit 31 (S2).

  When the storage of the target image is completed, the character string determination control unit 2b outputs a character string extraction command to the character string extraction unit 21.

  Upon receiving this character string extraction command, the character string extraction unit 21 extracts a target character string from the target image stored in the target document storage unit 31 (S3). The character string extraction unit 21 temporarily stores the extracted target character string in the target character string storage unit 32. Note that M is the number of extracted target character strings.

  When the extraction of the target character string is completed, the character string determination control unit 2b initializes a counter (not shown) provided therein to 1 (m = 1) (S4).

  Next, the character string determination control unit 2b determines whether or not the counter value (m) is equal to or less than the number M of character strings (S5).

  If the value (m) of the counter is not less than or equal to the number M of character strings (NO in S5), the character string determination control unit 2b ends a series of processes. That is, when all the target character strings extracted by the character string extraction unit 21 have been determined, the character string determination control unit 2b ends a series of processes.

  When the counter value (m) is equal to or less than the number M of character strings (YES in S5), character string determination control unit 2b outputs a character string pattern detection command to character string pattern detection unit 22a of specific character determination unit 22. To do.

  Upon receiving the character string pattern detection command, the character string pattern detection unit 22a refers to the character determination rule table 41 and the corresponding item table 42 to determine whether or not the mth target character string includes a character string pattern. Determine (specific character detection step). At this time, does the character string pattern detection unit 22a include the nth character string pattern (n is a number from 1 to the total number of character string patterns defined in the item correspondence table 42) of the item correspondence table 42? Whether or not is judged in order. The character string pattern detection unit 22a outputs this determination result to the character string determination control unit 2b. Details of the method for referring to the character determination rule table 41 and the item correspondence table 42 will be described later.

  When the character string pattern is not included (NO in S6), the character string pattern detection unit 22a outputs information indicating that to the character string determination control unit 2b. Upon receiving this information, the character string determination control unit 2b increases the counter value (m) by 1 (S10), and returns to S6.

  If a character string pattern is included (YES in S6), character string pattern detection unit 22a outputs the detected character string pattern to item determination unit 22b (S7).

  Upon receipt of this character string pattern, the item determination unit 22b refers to the corresponding item table 42 to determine a character string item candidate corresponding to the character string pattern (type determination step), and compares the determination result with an item comparison. To the unit 22c.

  Upon receiving this determination result, the item comparison unit 22c compares the plurality of determined item candidates with the recognition target item (the item selected in S1) stored in the target item storage unit 34, and compares the plurality of candidates. It is determined whether or not the recognition target item is included in the candidate (S8). The item comparison unit 22c outputs the determination result to the character string determination control unit 2b.

  When the recognition target item is not included in the plurality of candidates (NO in S8), the character string determination control unit 2b increases the counter value (m) by 1 (S11), and returns to S6.

  On the other hand, when the recognition target item is included in the plurality of candidates (YES in S7), the character string determination control unit 2b outputs a recognition command instructing to recognize the target character string to the character string recognition unit 23.

  When this recognition command is received, the character string recognition unit 23 refers to the character recognition dictionary corresponding to the recognition target item as described above, and specifies a specific character in the target character string stored in the target character string storage unit 32. Is determined (character recognition step) (S9). The character string recognition unit 23 stores the determination result (including the detected specific character) in the recognition result storage unit 34 and outputs completion information indicating that the storage is completed to the item determination unit 24.

  Upon receipt of this completion information, the item determination unit 24, when a specific character is included in the target character string stored in the recognition result storage unit 34, displays the item corresponding to the specific character as the target character string. (S10), and outputs the determination result to the character string determination control unit 2b.

  The character string determination control unit 2 b outputs the determination result to the display unit 8 via the output control unit 7.

  On the other hand, when a specific character is not included in the target character string, the item determination unit 24 outputs information indicating that to the character string determination control unit 2b. The character string determination control unit 2b receiving this information preferably does not output the determination result.

  When the item determination by the item determination unit 24 ends, the character string determination control unit 2b increases the counter value (m) by 1 (S11), and returns to S6.

  In the item determination unit 22b, if the item of the target character string is uniquely determined, the item determination process in S10 can be omitted. Further, when narrowing down a plurality of items to a unique item, it is preferable not to output the recognition result unless the item matches the recognition target item.

  Further, when a predetermined operation is performed by the user after the determination result is displayed on the display unit 8, the main control unit 2 reads the text information of the target character string consisting of the character code, that is, the character string recognition unit 23. The recognition result of the target character string obtained by the above may be stored in the telephone directory 44.

(Details of processing in specific character determination unit 22)
Next, details of processing in the specific character determination unit 22 will be described in more detail with reference to FIG. FIG. 13 is a flowchart showing the flow of processing in the specific character determination unit 22.

  When the character string pattern detection command is received, the character string pattern detection unit 22a first initializes a value (n) of a counter (not shown) provided therein to 1 (n = 1) (S21).

  Next, the character string pattern detection unit 22a refers to the value (n) of the counter and determines whether n is N1 or less (S22). Here, N1 is the number of pairs of character string patterns and items (the number of rows in the item correspondence table 42a) shown in the item correspondence table 42a in which one item can be specified. In the example shown in FIG.

  If n is N1 or less (YES in S22), the character string pattern detection unit 22a determines whether or not the target character string includes the nth character string pattern in the corresponding item table 42a (S23). ), And outputs the determination result and the detected character string pattern to the item determination unit 22b.

  When the target character string includes the nth character string pattern of the corresponding item table 42a (YES in S23), the item determination unit 22b has the item corresponding to the character string pattern as the item of the target character string. Is determined. Then, the item determination unit 22b outputs the determination result to the character string determination control unit 2b.

  When the target character string does not include the nth character string pattern in the corresponding item table 42a (NO in S23), the item determination unit 22b outputs the determination result to the character string pattern detection unit 22a. Then, the character string pattern detection unit 22a increases the value of n by 1 (S24), and returns to S22.

  On the other hand, if n is not N1 or less, that is, if the determination for all the character string patterns shown in item correspondence table 42a is completed (NO in S22), specific character determination unit 22 uses item correspondence table 42b. Is determined for the character string pattern shown in (S25 to S29). The processing after S25 is substantially the same as the processing up to S24, except that the determination is made for all character string patterns. This is because a character string pattern that cannot be limited to one item is detected.

  After the presence / absence of all the character string patterns is determined, the item determining unit 22b performs a process of narrowing down items from the item group corresponding to the character string patterns that have been detected so far (S29).

(Processing flow in the character string pattern detection unit 22a)
Next, a flow of processing for determining whether or not a character string pattern is included in the target character string will be described with reference to FIG. FIG. 14 is a flowchart showing the flow of processing in the character string pattern detection unit 22a.

  In the following description using the flowchart of FIG. 14, processing for determining whether or not the character string pattern matches the colon “:” + “//” will be described, but substantially the same processing is performed for other character string patterns. Is done.

  When the character string pattern detection command is received, the character string pattern detection unit 22a initializes a counter value (t) included in the character string pattern detection unit 22a to 1 (S31).

  Next, the character string pattern detection unit 22a determines whether t is T or less (S32). Note that T is the number of characters in the target character string.

  If t is larger than T, that is, if all the characters included in the target character string are determined, the process ends.

  If t is equal to or smaller than T, the character string pattern detection unit 22a determines whether or not the t-th character is a colon “:” (S33). The determination of whether or not it is a colon is as described above.

  When the t-th character is a colon (YES in S33), the character string pattern detection unit 22a determines whether or not the t + 1-th character is a character “//” composed of a series of slashes. (S34). If the slash is continuous (YES in S34), the character string pattern detection unit 22a ends the process, and if not (NO in S34), the process of S35 is performed.

  On the other hand, if the t-th character is not a colon (NO in S33), character string pattern detection unit 22a increments counter t by 1 (S35), and returns to the process of S32.

  The example shown in FIG. 14 is an example in which two slashes are treated as one character by overlapping with a vertical scanning line. If the two slashes do not overlap, it can be determined whether the t-th character is “:”, the t + 1-th character is “/”, and the t + 2-th character is “/”. .

  Whether or not other character string patterns are included can also be easily checked by the same method as in the flowchart of FIG.

(Example of processing when the document to be recognized is a business card)
Finally, how the business card shown in FIG. 3 is recognized by the processing described above will be described. It is assumed that the items of name, telephone, and mail are selected in S1 of FIG.

  The specific character string pattern described above is not included for each character string of the name “△ yama × m”, affiliation “general affairs department”, and company name “XYZ Corporation”. This is dealt with as follows.

  First, as described above, as described above, only the name may be extracted by the method described in Patent Document 2 and the like.

  There is no specific string pattern for the remaining affiliations and company names. For this, it is sufficient to determine whether the character string is a Japanese character string or an English character string by using a conventional method (for example, Patent Document 3). Since the affiliation and company name are determined to be Japanese character strings (should be), the recognition process should not be performed on these character strings.

  The remaining telephones, mails, and URLs can be specified by the above-described method for searching for a specific character string pattern.

"TEL: 0123-45-6789" on the telephone line is
(1) Since it matches the character string pattern (“L” + “:”), it is determined as a phone call, mail, or URL,
(2) Since it matches the character string pattern (hyphen + number string + hyphen), it is determined as a phone call or FAX.
Since the item common to (1) and (2) is only the telephone, the telephone line item is specified as “telephone”. Therefore, the telephone line is not subjected to character recognition processing in the character string recognition unit 23.

  Since “sankaku.batsu@sxyz.co.jp” in the mail line includes three periods, it is determined to be either mail or URL.

  In the above description of the present embodiment, since there is no rule that can identify which one, the mail line is determined to be either mail or URL, and is therefore a character recognition process target in the character string recognition unit 23.

  After the character string recognizing process in the character string recognizing unit 23, in the item determining process in the item determining unit 24, the target character string is determined to be a mail line due to the fact that the target character string has @, etc. 8 is displayed.

  Since “http://www.xyz.co.jp” in the URL line matches the character string pattern (“:” + “//”), it is determined as a URL. Therefore, the URL line is not subjected to character string recognition processing in the character string recognition unit 23.

  If the character string pattern cannot be extracted for some reason, it is determined whether the character string pattern is a Japanese character string or an English character string using the conventional method described in Patent Document 3 or the like. If so, it may be recognized as a telephone or mail item. Since telephone, FAX, mail, and URL are determined to be English character strings (should be), they may be recognized as telephone or mail items.

(Effect of mobile phone 1)
As described above, according to the mobile phone 1 (character string determination device) of the present embodiment, by focusing on the specific character string pattern included in the character string described in the recognition target document, the item of the character string is changed. It is possible to narrow down, and it is possible to uniquely determine a specific character string.

  Furthermore, since the item determination mainly uses at least one of the characteristics such as the character shape and the relative position of the character in the character string, the processing amount for the item determination is small. Therefore, the character string determination device is also suitable for mounting on a device with low processing capability such as a small portable information terminal.

  In the character string determination device of the present invention, only a specific item desired by the user can be extracted, and the recognition process can be applied to the specific item. This (and this method itself is also simple and fast) can reduce the recognition processing time even for devices with low processing capabilities such as small portable information terminals.

  In the character string determination device, there may be cases where the number of character string items is not always limited to one. For example, the determination result may be either mail or URL.

  However, there is an advantage that the speed of matching can be increased by limiting the items of character strings and using only necessary dictionary categories. If an item is not uniquely determined, a conventional method of determining an item from a character string recognition result after character string recognition may be used. Any conventional method may be used.

  Furthermore, the user can select items to be recognized, which is convenient for the user.

  Also, since only the necessary items selected by the user are recognized and the result is output, the burden on the user by deleting (or correcting) the items unnecessary for the user is greater than when all items are output. There is an advantage of less.

  In addition, even when the recognition target item is not limited, the character string determination device first narrows down the items of the character string, and performs a dictionary collation process using only the character category necessary for the narrowed target item, thereby recognizing the recognition target item. There is also an effect that the recognition processing time can be shortened rather than recognizing all items.

  The recognition target document is not limited to an image. For example, in the online handwritten character string recognition process, stroke information may be used instead of image information.

(Example of change)
The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope shown in the claims, and embodiments obtained by appropriately combining technical means disclosed in different embodiments. Is also included in the technical scope of the present invention.

  For example, in the above-described configuration, the input control unit 5 has received information on a recognition target item (specified type) that is an item of a target character string designated by the user, but information on the recognition target item is stored in advance in the storage unit 4. May be stored.

  Further, each block of the mobile phone 1 described above, in particular, the main control unit 2 may be configured by hardware logic, or may be realized by software using a CPU as follows.

  That is, the mobile phone 1 includes a CPU (central processing unit) that executes instructions of a control program that realizes each function, a ROM (read only memory) that stores the program, a RAM (random access memory) that expands the program, A storage device (recording medium) such as a memory for storing the program and various data is provided. The object of the present invention is to enable the computer to read the program code (execution format program, intermediate code program, source program) of the control program (character string recognition program) of the mobile phone 1 which is software for realizing the above-described functions. This can also be achieved by supplying the recorded recording medium to the mobile phone 1 and reading and executing the program code recorded on the recording medium by the computer (or CPU or MPU).

  Examples of the recording medium include tapes such as magnetic tapes and cassette tapes, magnetic disks such as floppy (registered trademark) disks / hard disks, and disks including optical disks such as CD-ROM / MO / MD / DVD / CD-R. Card system such as IC card, IC card (including memory card) / optical card, or semiconductor memory system such as mask ROM / EPROM / EEPROM / flash ROM.

  The mobile phone 1 may be configured to be connectable to a communication network, and the program code may be supplied via the communication network. The communication network is not particularly limited. For example, the Internet, intranet, extranet, LAN, ISDN, VAN, CATV communication network, virtual private network, telephone line network, mobile communication network, satellite communication. A net or the like is available. Also, the transmission medium constituting the communication network is not particularly limited. For example, even in the case of wired such as IEEE 1394, USB, power line carrier, cable TV line, telephone line, ADSL line, etc., infrared rays such as IrDA and remote control, Bluetooth ( (Registered trademark), 802.11 wireless, HDR, mobile phone network, satellite line, terrestrial digital network, and the like can also be used. The present invention can also be realized in the form of a computer data signal embedded in a carrier wave in which the program code is embodied by electronic transmission.

  The character string determination device of the present invention can determine which of the predetermined types the character string to be recognized corresponds to with a small amount of processing, and is therefore suitable for a portable terminal with low processing capability. Can be incorporated into.

It is a functional block diagram which shows the structure of the specific character determination part with which the mobile telephone of one Embodiment is provided. It is a functional block diagram which shows the structure of the said mobile phone. It is a figure which shows an example of the said recognition target document of the said mobile telephone. It is a figure which shows an example of the item of the target character string displayed on the display part with which the said mobile phone is provided. It is a figure which shows an example of a corresponding item table. It is a figure which shows an example of a corresponding item table. It is a figure which shows an example of a character determination rule table. It is a figure which shows the shape of a colon. It is a figure which shows the shape of a parenthesis. It is a figure which shows the shape of a slash. It is a figure which shows the example of the structure of the character string information extracted by the character string extraction part with which the said mobile phone is provided. It is a flowchart which shows the flow of the process in the said mobile telephone. It is a flowchart which shows the flow of a process in the specific character determination part with which the said mobile phone is provided. It is a flowchart which shows the flow of a process in the character string pattern detection part with which the said mobile phone is provided.

Explanation of symbols

1 Mobile phone (character string determination device)
22a Character string pattern detection unit (specific character detection means)
22b Item determination unit (type determination means)
23 Character string recognition unit (character recognition means)
24 Item determination part (character string specifying means)
43 character recognition dictionary 51 target item instruction section (accepting means)
60 Character string information (target character string)

Claims (8)

A character string determination device that acquires character string information indicating a shape of a character constituting a target character string, and determines a type of the target character string based on a character shape indicated by the character string information,
Specific character detection means for detecting whether or not a character string pattern, which is information indicating a specific shape and associated with a character string type, is included in the character string information;
When the character string pattern is detected in the character string information by the specific character detection means, a type determination means that sets the type of the character string corresponding to the character string pattern as the type of the target character string; A character string determination device comprising:   The character string determination according to claim 1, wherein the character string pattern includes information on a character having a specific shape, information on a character including a component having a specific relative positional relationship, or both. apparatus.   Character recognition means for identifying a character included in the character string information by referring to a plurality of character recognition dictionaries respectively corresponding to the type candidates when the type determination means lists a plurality of types of candidates. The character string determination device according to claim 1, further comprising:   When a plurality of candidates of the type are listed by the type determination means, and a target character string of a type that matches the specified type that is the type of the specified target character string is included in the plurality of candidates The apparatus according to claim 1, further comprising character recognition means for identifying a character included in the character string information by referring to a character recognition dictionary corresponding to the designated type. String judgment device.   The character string determination device according to claim 3 or 4, further comprising character string specifying means for specifying a type of the target character string based on a character recognition result of the character recognition means.   The character string determination program for functioning a computer as said each means of the character string determination apparatus of any one of Claims 1-5.   The computer-readable recording medium which recorded the character string determination program of Claim 6. A character string determination method in a character string determination device that acquires character string information indicating the shape of a character constituting a target character string and determines the type of the target character string based on the character shape indicated by the character string information. There,
A specific character detection step for detecting whether or not a character string pattern, which is information indicating a specific shape and associated with a character string type, is included in the character string information;
A type determining step in which, when the character string pattern is detected in the character string information by the specific character detecting means, the type of the character string corresponding to the character string pattern is set as the type of the target character string; A character string determination method comprising:

Images