JP5035848B2 - Item determination apparatus, item determination program, recording medium, and item determination method - Google Patents

Description

  The present invention relates to an item determination device, an item determination program, a recording medium, and an item determination method, and in particular, an item determination device, an item determination program for determining an item suitable for classification of character strings acquired from image data, The present invention relates to a recording medium and an item determination method.

  Currently, optical character reader (OCR) technology is used in various character readers such as document readers and form readers. Among them, as in a business card reader, a character string written on a business card is a character string corresponding to any item (attribute) such as first and last name (name), place name (address), company name (affiliation), etc. There are also devices that automatically determine whether or not.

  For example, in Japanese Patent Laid-Open No. 5-89292 (Patent Document 1), image data read from a scanner is converted into character code string data in a character recognition unit, and this data is converted into a character code keyword stored in a ROM. Alternatively, a character string recognizing device configured to classify while confirming a corresponding item while referring to dictionary data and output item-categorized character code string data DO including data specifying the item is disclosed. Yes.

  Japanese Laid-Open Patent Publication No. 59-47641 (Patent Document 2) describes a database creation device that identifies the largest printed character string of a character string printed on a business card as a first name and last name and inputs it to the database. Yes.

In recent years, the OCR has been miniaturized, and a camera-equipped mobile phone that takes a reading target with a camera and recognizes characters has been put on the market.
JP-A-5-89292 JP 59-47641 A

  However, OCR installed in a small information terminal such as a mobile phone generally has a specification that captures and recognizes an alphanumeric character string such as a mail address, URL (uniform resource locator), and telephone number. There are few small information terminals such as business card readers that have a function of reading Japanese character strings and automatically assigning them to items. As described above, one of the factors that the small information terminals having the function of reading Japanese characters and automatically sorting them into items is not so popular (not commercialized) is a large dictionary data capacity. Can be mentioned.

  For example, there is a business card reading device that uses a word dictionary (for example, a first name and last name dictionary, a place name dictionary, a general word dictionary) to determine a first name, a last name, and the like. This business card reader performs item classification by performing a process of collating the read characters with a dictionary. In this business card reading device, in addition to the character pattern matching dictionary essential for character recognition processing, a first name surname dictionary, place name dictionary, general word dictionary, etc. are required. Compared to small information terminals, which have a smaller program capacity and data size capacity, they are hindered.

For example, calculate the approximate size required to install 30000 “last name” dictionaries. Assume the average length of last names is 2 characters. Since a 2-byte storage capacity is required to represent a Japanese character code,
30000 (cases) x 2 (characters) x 2 (bytes) = 120 Kbytes
(1 Kbyte = 1000 bytes, and so on)
Storage capacity is required. Similarly, regarding the “name” dictionary, “place name” dictionary, and “general word” dictionary, when registering a practical number of words to some extent, a capacity of at least about several hundred Kbytes is required.

  This is because even if there are only a few hundred types of characters used, the number of types of words combined with them reaches tens of thousands or hundreds of thousands. Hereinafter, the “last name” dictionary, “first name” dictionary, “place name” dictionary, and “general word” dictionary are collectively referred to as a word dictionary. A general word is a common noun excluding proper nouns and pronouns.

  Although the capacity of a storage device has recently been increasing in recent small information terminals, in reality, the capacity of a storage device for software incorporation is not always sufficient due to multifunctionalization (that is, increase in installed software). For this reason, when it is mounted on a small information terminal, it is a problem to reduce the capacity of the word dictionary.

  In order to reduce the capacity of the word dictionary, for example, as described in JP-A-59-47641 (Patent Document 2), the largest printed character string of the character string printed on the business card is the first name and last name. There is a way to identify. If this method is used, it is not necessary to mount a first name surname dictionary in a small information terminal. Also, with regard to the determination of place names, if key characters and key words such as “prefectures”, “municipalities”, “chome, street address”, etc. are detected and determined as place names, the place name dictionary is not used. Place name item determination is possible.

  However, these determination methods that do not use the first name surname dictionary and place name dictionary have the following problems. In other words, regarding the determination of first and last names, the design of business cards is diverse, and the characters composed of the characters with the largest character size are not necessarily the first and last names.

  Some business card readers use keywords (~ company, ~ department, ~ head, etc.) to determine the company name, department name, and title. However, this business card reader has a problem in that item classification cannot be performed when a keyword is not found. For example, in determining a place name, a character string representing an address does not necessarily include key characters or key words such as “prefecture”, “city / town / ward”, and “chome, street address”.

  The present invention has been made to solve the above-mentioned problems, and the main object of the present invention is to make it possible to more appropriately determine items suitable for classification of inputted character strings while suppressing the capacity of the database. To provide a determination device, an item determination program, a recording medium, and an item determination method.

  According to an aspect of the present invention, an item determination device for determining an item of a character string includes, for each of at least one first partial character string included in the input character string, the first partial character. Recognizing means for recognizing each character included in the column, and reliability of each character determined in advance based on a first item in which at least one word including each character is to be classified is stored for each first item. And at least one character string of the first partial character string based on the reliability of each character included in the first partial character string by referring to the reliability table and the reliability table. Calculation means for calculating the reliability and first determination means for determining an item of the first partial character string based on the character string reliability.

  Preferably, the reliability of each character for each first item is determined based on the type of word including each character and the degree to which the word is used.

  Preferably, the recognition means acquires the position of each character included in the character string, divides the character string into at least one first partial character string based on the position of each character in the partial character string, and calculates The means refers to the reliability table, calculates the character string reliability based on the reliability of each character included in the first partial character string for each first item, and the first determination means includes: Based on the character string reliability for each first item, the first item of the first partial character string is determined.

  Preferably, the item determination device includes a first arrangement table that stores second items associated with arrangements of the plurality of first items, and a character string for each first item of the first partial character string. Second determination means for determining the second item of the character string based on the reliability and the arrangement of the first items is further provided.

  Preferably, the item determination device further includes a first arrangement table that stores second items associated with the arrangement of the plurality of first items. For each arrangement of the first items stored in the first arrangement table, the calculation means calculates the first partial character string based on the arrangement of the first items stored in the first arrangement table. Calculate string reliability. The first determination means determines the second item of the first partial character string based on the character string reliability for each arrangement of the first items stored in the first arrangement table.

  Preferably, the item determination device stores a word dictionary storing a plurality of words composed of at least one character belonging to any of the first items, and a first partial character string stored in the word dictionary. Judgment means for judging whether or not a word is included is further provided.

  Preferably, when the first partial character string includes a word stored in the word dictionary, the calculating means refers to the reliability table and is stored in the word dictionary of the first partial character string. Character string reliability for each first item of the second partial character string based on the reliability corresponding to each character included in the second partial character string composed of characters excluding the characters included in the word When the first partial character string does not include a word stored in the word dictionary, the reliability table is referred to, and each character included in the first partial character string is classified by the first item. Based on the reliability, at least one character string reliability of the first partial character string is calculated. The first determination means determines the first item of the second partial character string based on the character string reliability when the first partial character string includes a word stored in the word dictionary, When one partial character string does not include a word stored in the word dictionary, the first item of the first partial character string is determined based on the character string reliability.

  Preferably, the item determination device includes a first arrangement table that stores second items associated with the arrangement of the plurality of first items, and a word in which the first partial character string is stored in the word dictionary. And a third determination unit that determines the second item of the partial character string based on the first item to which the word of the word dictionary belongs and the first item of the second partial character string. Is further provided.

  Preferably, each of the first items is one of a last name, a first name, a place name, a prefecture name, a municipality name, and a general word.

  If another situation of this invention is followed, the item determination program for making a computer determine the item of a character string will be provided. The computer includes a reliability table in which the reliability of each character, which is determined in advance based on the first item in which at least one word including each character is to be classified, is stored for each first item. The item determination program recognizes each character included in the first partial character string for each of at least one first partial character string included in the input character string to the computer, and a reliability table A step of calculating at least one character string reliability of the first partial character string based on the reliability of each character included in the first partial character string for each first item; And determining an item of the first partial character string based on the column reliability.

  According to still another aspect of the present invention, a computer-readable recording medium recording an item determination program for causing a computer to determine an item of a character string is provided. The computer includes a reliability table in which the reliability of each character, which is determined in advance based on the first item in which at least one word including each character is to be classified, is stored for each first item. The item determination program recognizes each character included in the first partial character string for each of at least one first partial character string included in the input character string to the computer, and a reliability table A step of calculating at least one character string reliability of the first partial character string based on the reliability of each character included in the first partial character string for each first item; And determining an item of the first partial character string based on the column reliability.

  According to still another aspect of the present invention, an item determination method using an item determination device for determining an item of a character string is provided. The item determination device includes a reliability table in which the reliability of each character determined in advance based on a first item in which at least one word including each character is to be classified is stored for each first item, and a control device With. In the item determination method, the control device recognizes each character included in the first partial character string for each of at least one first partial character string included in the input character string; Calculates at least one character string reliability of the first partial character string based on the reliability of each character included in the first partial character string by the first item with reference to the reliability table And a step of determining the item of the first partial character string based on the character string reliability.

  As described above, according to the present invention, an item determination device, an item determination program, a recording medium, and an item determination method that can more appropriately determine an item suitable for classification of an input character string while suppressing the capacity of the database. Can be provided.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description, the same parts are denoted by the same reference numerals, and when the names and functions of the parts are the same, detailed description of the parts will not be repeated.

[Embodiment 1]
<Overall configuration>
First, the overall configuration of the item (attribute) determination device according to the present embodiment will be described. The item determination apparatus according to the present embodiment is an apparatus for determining an item of a recognized character string, and is typically a small information terminal such as a camera-equipped mobile phone 100 or a PDA, a personal computer, or a workstation. Realized by a computer. The item determination process performed by the item determination apparatus is executed by a control device of a small information terminal such as a camera-equipped mobile phone 100 or a PDA or a control device of a computer such as a personal computer or a workstation. Is realized.

  FIG. 1 is an external perspective view showing a camera-equipped mobile phone 100 according to the present embodiment. FIG. 1 shows a foldable camera-equipped mobile phone 100 as a small terminal device, but is not limited to such a form. FIG. 1 is a front perspective view of the cellular phone 100 with the casings 114 and 116 opened.

  As shown in FIG. 1, the mobile phone 100 includes a first casing 114, a second casing 116, a first speaker 173, a second speaker 174, a main display (monitor 150), A first camera 142, a second camera 143 (see FIG. 2), an input unit 120, and a microphone 172 are included. The first casing 114 and the second casing 116 are foldably connected at the connection portion 115.

  The input unit 120 includes a function key group 122 including a plurality of keys such as an enter key 122A, and a numeric key group 124 including a numeric keypad for inputting numbers and the like. The function key group 122 moves at least a part of the display on the main display (monitor 150) (the entire display or a cursor displayed on the main display (monitor 150)) by a predetermined distance vertically and horizontally. For example, an upper key 121A, a lower key 121B, a left key 121C, a right key 121D, and a determination key 122A for inputting that the processing corresponding to the moved cursor or the like is determined.

<Hardware configuration>
Next, a hardware configuration of mobile phone 100 according to the present embodiment will be described. FIG. 2 is a block diagram showing a hardware configuration of mobile phone 100 as an example of the communication apparatus according to the present embodiment. As shown in FIG. 2, mobile phone 100 according to the present embodiment includes antenna 1, radio communication unit 101 that receives an input of a radio signal received by antenna 1, a first memory 106, a second memory A memory 107, an input unit 120 that receives input of various instructions from the outside, a control device 105 for controlling each operation of the mobile phone 100, a monitor 150 for displaying an image, and an external audio input And a speaker 173 and 174 for outputting sound based on an electric signal (audio signal).

  The wireless communication unit 101 transmits a wireless signal via the antenna 1 and receives a wireless signal from the outside according to control by the control device 105.

  The first camera 142 is a camera provided inside the first housing 114, and is for taking a picture of a user of the mobile phone 100 when making a videophone call, for example. The second camera 143 is a camera provided on the outside of the first housing 116 and is used for taking a picture of a landscape or taking an image on a document such as a barcode. is there. Image data of images taken by the first camera 142 and the second camera 143 is sent to the control device 105, and the control device 105 stores the image data in the first memory 106 or the second memory 107.

  As shown in FIGS. 1 and 2, the mobile phone 100 according to the present embodiment takes an image on a document such as a business card 202 with a second camera 143 provided outside the first housing 114. The image data of the image can be stored in the first memory 106 or the second memory 107.

  The control device 105 is realized by, for example, a CPU (Central Processing Unit) and other arithmetic processing devices and programs for realizing various functions described later. The control device 105 and other components are Connected via signal line. More specifically, in the above-described various processes, for example, programs related to various processing procedures stored in the second memory 107 are once read to the first memory 106, and the read programs are stored in the control device 105. This is realized by being executed in step (b). However, the control device 105 may be realized by a dedicated hardware circuit that realizes each function.

  The control device 105 realizes a call function of the mobile phone 100 while linking with other components such as the wireless communication unit 101. The control device 105 controls each element of the mobile phone 100 and performs various calculations. In particular, the control device 105 performs an item determination process as will be described later.

  The input unit 120 is realized by a function key group 122 including a plurality of keys such as a determination key 122A and a numeric key group 124 including a numeric keypad for inputting numbers and the like, and receives information from a user by key input or the like.

  The monitor 150 includes a liquid crystal panel and a CRT, and displays information output from the control device 105.

  The first memory 106 is used as a work RAM (Random Access Memory). The first memory 106 is generally a volatile memory, but may be a writable nonvolatile memory. More specifically, the first memory 106 according to the present embodiment typically includes a DRAM (Dynamic Random Access Memory), an SRAM (Static Random Access Memory), a PSRAM (Pseudo-SRAM), an SDRAM ( Synchronous DRAM) etc.

  The second memory 107 stores data such as a control program. The second memory 107 is generally a non-volatile memory. The second memory 107 is typically realized by an EEPROM (Electrically EPROM), a NAND flash memory, a NOR flash memory, a hard disk, or the like. Data other than the control program (font data, adjustment data, user settings, etc.) is mainly stored in the second memory 107, but may be temporarily stored in the first memory 106. Good.

  The first speaker 173 is for outputting call voice and is provided inside the housing 114. On the other hand, the second speaker 174 outputs an incoming call voice or a call voice during a hands-free call, and is provided on the side surface of the housing 116, for example.

  The item determination apparatus and the item determination process according to the present embodiment are realized by hardware such as a mobile phone 100 with a camera and software such as a control program. Generally, such software is stored in the second memory 107 at the time of manufacture or distributed by a user downloading it via a network or the like. Then, the software is read from the second memory 107 to the first memory 106 and executed by the control device 105.

<Functional configuration>
FIG. 3 is a block diagram illustrating a functional configuration of the mobile phone 100 according to the first embodiment. As illustrated in FIG. 3, the mobile phone 100 includes a photographing unit 142-1, an extraction unit 105-2, a recognition unit 105-3, a calculation unit 105-4, a first determination unit 105-5, Second determination unit 105-6, output unit 105-6, image storage unit 107-2, character recognition dictionary 107-3, result storage unit 107-4, reliability table 107-5, and arrangement Table 107-6.

  Here, the extraction unit 105-2, the recognition unit 105-3, the calculation unit 105-4, the first determination unit 105-5, and the second determination unit 105-7 are, for example, a control device such as a CPU. 105 is realized by reading the control program (software) stored in the second memory 107 into the first memory 106 and executing the control program (software). The image storage unit 107-2, the character recognition dictionary 107-3, the result storage unit 107-4, the reliability table 107-5, and the arrangement table 107-6 are, for example, the first memory 106 or the second memory 106. This is a function included in the storage unit 107-1 realized by the memory 107.

  Hereinafter, various functions of the mobile phone 100 will be described. As described above, in the function of each block and the processing of each step described below, the control device 105 reads out the control program (software) stored in the second memory 107 to the first memory 106 and executes the control program ( (Software) may be implemented, or each may be implemented by a dedicated hardware circuit or the like.

  The photographing unit 142-1 is realized by, for example, the first camera 142, the second camera 143, or a scanner (not shown). It is to convert to data. The photographing unit 142-1 temporarily stores the acquired image data in the image storage unit 107-2. More specifically, the image data acquired by the imaging unit 142-1 is temporarily stored in the volatile first memory 106 or the like, and then stored in the second memory 107 in accordance with an external save command or the like. Remembered.

  The extraction unit 105-2 reads image data from the image storage unit 107-2 and extracts a character string from the image data. More specifically, the extracting unit 105-2 extracts a character string (such as character string position coordinate information) as an image from the image data temporarily stored in the image storage unit 107-2. The extraction result by the extraction unit 105-2 is the circumscribed frame (upper left and lower right) coordinates of the black pixel (character pixel) connected component constituting each character string. Here, the extraction result by the extraction unit 105-2 may be temporarily stored in the result storage unit 107-4 of the storage unit 107-1.

  However, the extraction of the character string by the extraction unit 105-2 is not limited to the extraction of the character string from the image data. For example, the extraction unit 105-2 acquires character string data read from a business card, address book, or the like by an OCR that does not have an item determination function from another device, a recording medium, another application (control program), or the like. It may be a thing. For example, the extraction unit 105-2 may generate a character string based on the input voice data, or may acquire a candidate string for conversion to sign language or conversion to kanji. For example, the extraction unit 105-2 may acquire a character string by another acquisition method.

  The recognition unit 105-3 recognizes at least one first partial character string included in the input (acquired) character string and a character included in the first partial character string. More specifically, the recognition unit 105-3 acquires the position of each character included in the character string, and divides the character string into at least one first partial character string based on the position of each character. More specifically, the recognizing unit 105-3 refers to the character string formed of the character code information while referring to the character recognition dictionary 107-3 for the character string as the image information extracted by the extracting unit 105-2. Is generated. Here, the character string recognition result by the recognition unit 105-3 includes character code information (may include recognition candidates) of each character determined by the recognition unit 105-3, coordinate information for each character, and the like. is there. The character recognition dictionary 107-3 is a standard pattern dictionary for character recognition used in the recognition unit 105-3.

  FIG. 4 is an image diagram showing a data structure of a character string recognition result database stored in the result storage unit 107-4. As shown in FIG. 4, the information stored in the character string result storage unit 107-4 includes the number L of character strings in the input image and the number of characters constituting the character string in the nth row (1 ≦ n ≦ characters). (Number of lines L), (circumscribed frame) coordinates of each character of the character string of the nth line, character code, number of character string separation positions, and each separation position (information indicating coordinates, or what is counted from the top) For example, information indicating whether it is after the second character).

  As described above, the recognition unit 105-3 separates the character string into several first partial character strings from the character string recognition result stored in the result storage unit 107-4. As a separation method for separating character strings into partial character strings, for example, after calculating the average distance from the distance between each character, the average distance is larger than the average distance (the distance between the characters being noticed is the average distance × predetermined) If the threshold value is greater than or equal to the threshold value, a method of generating a partial character string by dividing the character string at that position is exemplified. The character string separation result (information indicating the partial character string) by the recognition unit 105-3 may be recorded in any form, but in this embodiment, the result storage unit 107-4 shown in FIG. As described above, by storing the number of separation positions and the respective separation positions, the character string separation result is recorded.

  Here, FIG. 5 is an image diagram showing a character string and a partial character string in the image data. As illustrated in FIG. 5, the recognition unit 105-3 recognizes a character string or a partial character string from the image data read from the business card 201. For example, after recognizing the character string "Taro Hayakawa", the character string "Hayakawa Taro" is replaced with the character string "Hayakawa" and the part "Taro" based on the spacing between the characters that make up "Taro Hayakawa". They are separated into character strings. Then, the recognition result is stored in the result storage unit 107-4.

  When the extraction unit 105-2 acquires character string data based on the voice data, the recognition unit 105-3 separates the partial character string from the character string based on the time interval at which each character is input. . For example, after calculating the average time interval from the time interval at which the voice corresponding to each character is input, it is larger than the average time interval (the time interval of interest is equal to or greater than the average time interval x a predetermined threshold) In this case, a method of generating a partial character string by dividing a character string at that position can be used.

  And the recognition part 105-3 can recognize a partial character string from a character string based on the space in the input character string data.

  In the reliability table 107-5, the reliability of each character determined based on an item in which at least one character string including each character is to be classified is stored for each first item. Then, the reliability of each character by the first item is determined based on the degree to which at least one character string including each character is used and the item into which the character string is to be classified.

  FIG. 6 is an image diagram showing a data structure of the reliability table 107-5. As shown in FIG. 6, the reliability table 107-5 includes, for each character (for each character code), reliability when used for “last name”, reliability when used for “first name”, and “place name”. And the reliability when used for “general word” are stored (in FIG. 6, “general word” is described as “general”).

<Reliability table>
Here, a method of creating a reliability table will be described. As shown in FIG. 6, the reliability table 107-5 shows how much is used as a surname, first name, place name, and general word for each kanji character stored in the reliability table. The degree (use degree) is recorded. A specific method for creating the reliability table 107-5 may be as follows, for example.

  For surnames and surnames, a database (corpus data) of Japanese surnames and surnames is used. FIG. 7 is an image diagram showing an example of a Japanese surname database. In this database, Japanese surnames are arranged in descending order of the population with the last names. The rightmost column shows the ratio of people with this surname per 10,000 people. The population ratio value of the last name in the rightmost column is added to one kanji character used for each surname (Sato, Suzuki, ...) in this database.

For example, the three surnames Sato, Suzuki, and Naito will be described. Since the proportion of the population with Sato's surname is 141, the proportion of the population with Suzuki's surname is 137, and the proportion of the population with Naito's surname is 16, the degree of use of each letter considering only these three surnames Is
Sa ... 141
Wisteria ... 157 (141 + 16)
Bell ... 137
Tree ... 137
Within ... 16
It is expressed. Since the sum of the numerical values is 588 (141 + 157 + 137 + 137 + 16), dividing each numerical value by 588 and then multiplying by 100 will normalize the total to 100 (%). The normalized value is
Sa ... 23.97
Wisteria ... 26.70
Bell ... 23.29
Thu ... 23.29
Inside ... 2.72
It becomes.

  The numerical value of each item for each character is conveniently stored in 1 byte (2 bytes) as described above. In the case of 1 byte, the use level of each character is divided by 588 and multiplied by 256. It may be configured to be a numerical value between 0 and 255 by rounding down the first decimal place. Thereby, the memory capacity of 1 byte can be suppressed for each item of each character.

  Alternatively, the degree of use of each character for each item may be normalized by setting the largest numerical value of all characters and all items to 255. In other words, the degree of use of each character for each item may be divided by the largest numerical value and multiplied by 255.

  Furthermore, for each item, the largest numerical value of all characters may be set to 255, and the degree of use of each character for each item may be normalized. In other words, the degree of use of each character for each item may be divided by the largest numerical value and multiplied by 255.

  If the method as described above is used, it is possible to create a reliability table in which the reliability for each item of the “last name” is stored for all characters handled by the mobile phone 100. However, as in the database shown in FIG. 7, it is not always possible to obtain the ratio of the population of each surname. For example, if only rank information (in order of descending number of surnames) is known, each surname may be weighted according to the rank, and the degree used as the last name of each character may be calculated. Furthermore, when even the order is unknown, all the weights may be calculated as 1.

  Then, the reliability of the “name”, “place name”, and “general” items of each character can be created in the same manner as in the case of the “last name” described above. As for the reliability of the item “place name” of each character, the place name may be weighted according to the population of the place name, and the reliability of the characters constituting the place name may be calculated. For “general words” (words other than surnames, first names, place names), for example, words handled by the mobile phone 100 are extracted in advance, the weight is constant, and the reliability of each character according to the type of the word May be calculated.

  In the present embodiment, the example of creating the character by item is limited to kanji, but hiragana and katakana may also be included.

  Although the reliability table has been described above, the method of creating the reliability table is not limited to the above method. That is, if the reliability of each character is determined based on the type of word including each character, the appearance frequency of the word, etc., and the reliability of each character determined in this way is stored so that it can be referred to Good. For example, since it is only necessary to be able to acquire the difference (characteristics, attributes) of each character between last name, first name, place name, and general, whether a character string including each character is entered in a dictionary of a specific specialized field It is also effective to determine the degree of reliability accordingly, and to determine the magnification (weighting) according to the type of the dictionary (for example, a rare name dictionary, a rare place name dictionary, etc.).

<Functional configuration>
The calculation unit 105-4 refers to the reliability table, and determines at least one character of the first partial character string based on the reliability of each character included in the first partial character string for each first item. The column reliability is calculated and stored in the first memory 106 or the like. The calculation unit 105-1 according to the present embodiment refers to the reliability table 107-5, and sets the character string based on the reliability of each character included in the first partial character string for each first item. Calculate reliability.

  The first determination unit 105-5 determines the first item of the first partial character string based on the character string reliability. More specifically, the first determination unit 105-5 determines the first item of the first partial character string based on the character string reliability for each first item.

  The output unit 105-7 is realized not only by the control device 105 but also by the monitor 150, a peripheral I / F circuit (not shown), and the like, and various kinds of information such as a recognition result stored in the result storage unit 122 are determined. The first item and the second item are transferred from the item determination program to another application or output to the outside of the mobile phone 100.

  The arrangement table 107-6 stores the second item in association with the arrangement of the first item. FIG. 8 is an image diagram showing a data structure of the arrangement table. As shown in FIG. 8, in the case of a name having a large character interval, the arrangement table shows whether or not the arrangement of the items is appropriate for the entire character string after one item is associated with each partial character string. In addition, if it is valid, it is a table for determining to which item it is appropriate to classify the entire character string.

  In FIG. 8, only combinations that are appropriate as the arrangement of items are listed for character strings including two to four partial character strings. For example, in the case of a character string composed of three partial character strings, the first partial character string is “last name”, the second partial character string is “last name”, and the third partial character string is “name”. The arrangement of the partial character strings is stored in the arrangement table as valid. The column to the right of “⇒” in FIG. 8 indicates that the character strings in the arrangement of the partial character strings are suitable for being classified into the item “first and last name” as a whole.

  On the other hand, as shown in FIG. 8, for example, a combination in which the first partial character string is “first name”, the second partial character string is “last name”, and the third partial character string is “first name” is a partial character. It is not stored in the arrangement table 107-6 because it is not appropriate as the arrangement of columns.

  Here, as the first item, for example, “last name”, “first name”, “place name”, “prefectural name”, “city name”, “general word”, “title”, “department”, etc. Can be mentioned. The second item includes “first name (name)”, “address (place name)”, “general word”, “company name”, “title”, “affiliation”, and the like.

  The second determination unit 105-6 determines the second item of the character string based on the item for each first partial character string and the arrangement method stored in the arrangement table 107-6. More specifically, the second determination unit 105-6 determines the character string reliability for each first item of the first partial character string and the arrangement method of the first items stored in the arrangement table 107-6. Based on the above, the second item of the entire character string is determined.

<Item judgment processing>
Next, an item determination process according to the present embodiment will be described. FIG. 9 is a flowchart showing a processing procedure of item determination processing in the mobile phone 100 according to the first embodiment. FIG. 10 is an image diagram showing a business card 202 that is a target of the item determination process according to the first embodiment. Here, the item determination process when “Taro Hayakawa” is input among the character string recognition results extracted by the control device 105 and recognized by the control device 105 will be described. It is assumed that the character string recognition result is already stored in the storage unit 107-1 realized by the first memory 106 or the second memory 107.

  In the case of a business card 202 as shown in FIG. 10, the surname and first name “Taro Hayakawa” are not written large in the center of the business card 202. Therefore, in the conventional method, the surname, first name, place name, general name It is not easy to determine that “Taro Hayakawa” is a surname without using various word dictionaries such as words.

  First, the control device 105 converts the recognition result character string “Taro Hayakawa” into a partial character string “Hayakawa” and a partial character string “Taro” based on the character string recognition result stored in the storage unit 107-1. (Step S101). As described above, the method of separating the character strings can be easily performed by a method of comparing the distance between the characters and the average of the distance between the characters. Assuming that the number of partial character strings after separating the character strings is N, in the case of “Taro Hayakawa”, N = 2.

  In this description, the processing when the control device 105 separates the recognized character string into a plurality of parts is described. However, when the control device 105 cannot separate the recognized character string because the distance between characters is constant. The following processing is performed assuming that the character string includes one partial character string (assuming that there is only one partial character string).

  In the first place, when the character string itself obtained by the control device 105 is handled as one partial character string, the extracted character string itself is handled as a partial character string. In this case, the recognition unit 105-3 realized by the control device 105 recognizes a character string and characters constituting the character string.

  The control device 105 initializes the counter n to 1 (step S102). If n ≦ N (YES in step S103), control device 105 calculates the reliability of the nth partial character string and the item to be classified (step S104 to step S105).

  The control device 105 obtains the character string reliability of the nth partial character string for each item. The processing for obtaining the character string reliability may be performed by adding the reliability of each character in the reliability table defined for each character and each item as shown in FIG. S104).

FIG. 11 is an image diagram showing the reliability of the characters constituting the first partial character string and the character string reliability of the first partial character string according to the first embodiment. As shown in FIG. 11, for each partial character string (here, the first partial character string “Hayakawa”), the reliability of each character is added for each item. That is, the character string reliability for each item of the first partial character string “Hayakawa” is
(Last name) ... 11
(Name) ... 5
(Place name) ・ 10
(General) ... 3
It becomes.

  The control device 105 determines that the first item having the maximum character string reliability is the first item suitable for classification of the first partial character string (step S105). In this case, it is determined that the first partial character string is a partial character string to be classified into the item “last name”.

  Then, the control device 105 increments the counter (n = n + 1) (step S106). Since n ≦ N (YES in step S103), control device 105 calculates the reliability of the second partial character string (here “Taro”) and the item to be classified (step S104). -Step S105).

  That is, the control apparatus 105 calculates | requires the character string reliability of the 2nd partial character string for every item. Similarly to the above, the control device 105 adds the reliability of each character in the reliability table for each item (step S104).

FIG. 12 is an image diagram showing the reliability of the characters constituting the second partial character string and the character string reliability of the second partial character string according to the first embodiment. As shown in FIG. 12, the character string reliability for each item of the second partial character string “Taro”, calculated by adding the reliability of each character for each item,
(Last name) ... 4
(Name) ... 21
(Place name) ・ ・ 9
(General) ... 5
It is.

  The control device 105 determines that the first item having the maximum character string reliability is the first item suitable for classification of the second partial character string (step S105). In this case, the control device 105 determines that the second partial character string is a partial character string that should be classified into the item “name”. Then, the control device 105 increments the counter (n = n + 1) (step S106).

  As described above, in steps S104 to S105, the control device 105 determines the item “last name” that is the highest character string reliability for the first character string “Hayakawa” and the second character string “Taro”. The item of “name” having the highest character string reliability is determined.

  On the other hand, if n> N (NO in step S103), control device 105 determines the second item of the entire character string from the combination (arrangement) of the partial character strings determined in step S106. (Step S107). For this, by using the arrangement table 107-6 as shown in FIG. 8, the control device 105 determines to which item a character string composed of at least one partial character string should be classified. Specifically, as shown in FIG. 8, the second item “first name” corresponds to the arrangement of the first item “first name” next to the first item “last name”. In this way, it can be determined that “Taro Hayakawa” is a first and last name.

  When the arrangement of items with respect to the partial character string is an arrangement that is not stored in the arrangement table 107-6, for example, the control device 105 performs the following calculation. That is, the control device 105 selects the first item of the partial character string having the next highest reliability as the reliability of the entire character string. Then, the control device 105 searches whether the arrangement of the first items is stored in the arrangement table 107-6, and when the arrangement of the first items is stored in the arrangement table 107-6. The second item as the entire character string is determined.

  In the present embodiment, only the case where N = 2 has been described. Of course, the case where N is 1 or N is 3 or more is also possible. However, as N increases, the amount of processing increases due to an increase in the number of combinations. If the number of characters is large, it may be limited to a small number of characters, such as narrowing down to the first few characters.

[Embodiment 2]
Next, a second embodiment of the item determination device will be described. Also in the present embodiment, a mobile phone 100 will be described as an example of an item determination device. Mobile phone 100 in the present embodiment is different from mobile phone 100 in the first embodiment only in a part of the functional configuration and the processing procedure in the item determination process. That is, the overall configuration and hardware configuration of mobile phone 100 in the present embodiment is the same as mobile phone 100 in Embodiment 1 shown in FIGS. 1 and 2, and therefore description thereof will not be repeated here.

  The item determination device and the item determination processing according to the present embodiment are also realized by hardware such as the mobile phone 100 with a camera and software such as a control program. In general, such software is stored in the second memory 107 at the time of manufacture, or distributed by a user downloading it via a network or the like. Then, the software is read from the second memory 107 to the first memory 106 and executed by the control device 105.

<Functional configuration>
FIG. 13 is a block diagram showing a functional configuration of mobile phone 100 according to the present embodiment. As illustrated in FIG. 13, the mobile phone 100 includes a photographing unit 142-1, an extracting unit 105-2, a recognizing unit 105-3, a calculating unit 105-4, a first determining unit 105-5, It includes an output unit 105-7, an image storage unit 107-2, a character recognition dictionary 107-3, a result storage unit 107-4, a reliability table 107-5, and an arrangement table 107-6.

  Here, the extraction unit 105-2, the recognition unit 105-3, the calculation unit 105-4, the first determination unit 105-5, and the output unit 105-7 are, for example, a control device 105 such as a CPU. This is realized by reading the control program (software) stored in the second memory 107 into the first memory 106 and executing the control program (software). The image storage unit 107-2, the character recognition dictionary 107-3, the result storage unit 107-4, the reliability table 107-5, and the arrangement table 107-6 are functions included in the storage unit 107-1. is there. The storage unit 107-1 is realized by the first memory 106 or the second memory 107.

  Hereinafter, various functions of the mobile phone 100 will be described. As described above, the functions of the following blocks and the processing of each step are performed by a control program (software) stored in the second memory 107 by the control device 105. May be realized by reading out to the first memory 106 and executing the control program (software), or each may be realized by a dedicated hardware circuit or the like.

  It should be noted that the imaging unit 142-1, the extraction unit 105-2, the image storage unit 107-2, the character recognition dictionary 107-3, the result storage unit 107-4, and the reliability table 107- in the present embodiment. 5 has the same configuration as that of the first embodiment, and therefore, description thereof will not be repeated here. Also, the method of generating the reliability table 107-5 in the present embodiment has the same configuration as that in the first embodiment, and therefore description thereof will not be repeated here.

  First, as shown in FIG. 8, the arrangement table 107-6 stores the second item in association with the arrangement of the first items. The arrangement table according to the present embodiment stores the arrangement of items corresponding to each character, and is a table for determining what items should be classified as the entire partial character string based on the arrangement. .

  As shown in FIG. 8, the arrangement table 107-6 stores only arrangements that are appropriate as arrangements of items corresponding to each character for partial character strings including two to four characters. For example, in the case of a partial character string consisting of three characters, the first character is “last name”, the second character is “last name”, and the third character is “first name”. Stored in a table. Note that the right column of “⇒” in FIG. 8 indicates that the partial character strings in the arrangement are suitable for being classified into the item “first and last name” as a whole.

  On the other hand, in FIG. 8, for example, a combination in which the first character is “first name”, the second character is “last name”, and the third character is “first name” is not valid as a normal character string. , Not stored in the arrangement table 107-6.

  The recognition unit 105-3 recognizes at least one first partial character string included in the character string and a character included in the first partial character string. Then, the recognition unit 105-3 acquires the position of each character included in the character string, and divides the character string into at least one first partial character string based on the position of each character. However, it is assumed that the recognition unit 105-1 according to the present embodiment handles a character string as one partial character string.

  More specifically, the recognition unit 105-3 generates a character string as character code information by referring to the character recognition dictionary 107-3 with respect to the character string as image information extracted by the extraction unit 105-2. To do. Here, the character string recognition result by the recognition unit 105-3 is character code information (may include recognition candidates) of each character determined by the recognition unit 105-3, coordinate information of each character, and the like. .

  The calculation unit 105-4 refers to the reliability table, and determines at least one character of the first partial character string based on the reliability of each character included in the first partial character string for each first item. The column reliability is calculated and stored in the first memory 106 or the like. The calculation unit 105-1 according to the present embodiment is based on the arrangement of the first items stored in the arrangement table 107-6 for each arrangement of the first items stored in the arrangement table 107-6. Thus, the character string reliability of the first partial character string is calculated. In other words, the calculation unit 105-1 calculates the reliability of the partial character string for each arrangement of only the items in which the arrangement of the characters constituting the partial character string is valid as the arrangement of the first item. To do.

  The first determination unit 105-5 determines the first item of the first partial character string based on the character string reliability. More specifically, the first determination unit 105-5 determines the second item of the first partial character string based on the character string reliability for each arrangement method of the first items stored in the arrangement table. judge.

That is, as shown in FIG. 8, when the character string recognized by the recognition unit 105-3 is composed of three characters, the calculation unit 105-4 refers to the arrangement table 107-6 and character strings composed of three characters. The following 6 types are extracted. That is,
Last name + surname + surname + surname + surname + single + single + single + single + ground + single + single + single.

  The calculation unit 105-4 calculates the character string reliability of the partial character string for each of the arrangement methods described above, and determines the second character string corresponding to the arrangement method having the maximum character string reliability as the second character string second. Judge as item.

  The output unit 105-7 is realized not only by the control device 105 but also by the monitor 150, a peripheral I / F circuit (not shown), and the like, and various kinds of information such as a recognition result stored in the result storage unit 122 are determined. The second item is transferred from the item determination program to another application or output to the outside of the mobile phone 100.

<Item judgment processing>
Next, an item determination process according to the present embodiment will be described. FIG. 14 is a flowchart showing a processing procedure of item determination processing in mobile phone 100 according to the present embodiment.

  Hereinafter, a processing procedure of item determination processing in the mobile phone 100 that handles an input character string as one partial character string will be described. However, the recognition unit 105-3 realized by the control device 105 treats the input character string as being composed of a partial character string consisting of one character, that is, N when a character string consisting of N characters is input. It may be configured to be separated into individual partial character strings.

  Here, the case where the image data of the business card 202 as shown in FIG. 10 is input will be described. The target character string is “South Sone”. In the present embodiment, the description will be made on the assumption that the character string includes one partial character string “Minami Sone”. However, as described above, the character string separation number N may be three. . In the present embodiment, as shown in FIG. 5, in addition to the partial character string “Minami Sone”, a partial character string “Hayakawa Kogyo”, a partial character string “Taro”, and the like are extracted.

  Referring to FIG. 14, first, based on the character string recognition result stored in the storage unit 107-1, the control device 105 converts the recognition result character string “Nan Sone” into one part “Nan Sone”. Recognized as a character string (step S201).

  The control device 105 sequentially extracts the arrangement for three characters from the arrangement table shown in FIG. 8 according to the number of characters included in the partial character string (step S202). However, in step S202, only the number of arrangement methods may be extracted, and the nth arrangement method corresponding to n may be extracted in step S205. Then, the control device 105 initializes the counter n to 1 with N being the total number of arrangements corresponding to the number of characters, corresponding to the counter for each arrangement method (step S203).

That is, as shown in FIG. 8, when the partial character string is composed of three characters, the control device 105 uses the arrangement table 107-6 as the arrangement method of the first item corresponding to the character string consisting of three characters. The following 6 types are extracted. That is, the control device 105
Last name + surname + surname + surname + surname + single + single + single + single + ground + single + single + single, N = 6, partial character string for each arrangement The character string reliability corresponding to is calculated. However, as described above, in step S202, only the number of arrangements may be extracted, and the arrangement corresponding to n may be extracted in step S205.

  If n ≦ N (YES in step S204), control device 105 calculates the character string reliability of the partial character string and the item to be classified for the n-th arrangement (step S205). That is, in step S205, the character string reliability is calculated only for combinations that are valid as the arrangement method of the partial character strings. In other words, in step S205, the control device 105 calculates, for each arrangement, the character string reliability of the partial character string for the nth arrangement, and the storage unit 107- together with the second item corresponding to the arrangement. 1 is stored. When calculating the character string reliability for one arrangement, the control device 105 increments the counter (n = n + 1) (step S206).

  On the other hand, if n> N (NO in step S204), control device 105 determines the second item of the partial character string from the arrangement of the partial character strings for each arrangement determined in step S205. (Step S207). Specifically, the control device 105 reads out the character string reliability of the partial character strings for each arrangement and the second item corresponding to the arrangement from the storage unit 107-1, and obtains the highest character string reliability. Based on the arrangement, the second item to be classified is determined.

  Hereinafter, the case where “South Sone” is extracted will be described in more detail. FIG. 15 is an image diagram showing the reliability of characters constituting the partial character string according to the second embodiment. Referring to FIG. 8, control device 105 extracts six types of arrangement as possible arrangements of a three-character string (step S202). The control device 105 obtains the character string reliability for each of the six types of arrangement (step S205).

More specifically, as described below, the control device 105 calculates the string reliability for the six types of first item arrangement, and the storage unit 107 together with the second item corresponding to each arrangement. -1.
Last name + Last name + First name = 6 + 4 + 1 = 11⇒ "Last name"
First name + First name + First name = 6 + 1 + 1 = 8⇒ "Last name"
Last name + Single + Single = 6 + 1 + 1 = 8⇒ "Word"
Earth + Single + Single = 6 + 1 + 1 = 8⇒ "Word"
Place + place + place = 6 + 4 + 3 = 13⇒ "place name"
Single + single + single = 1 + 1 + 1 = 3 =>"word"
It becomes.

  The control device 105 selects place + place + place as the arrangement of the first items having the maximum character string reliability, and the character string is classified as “place name” corresponding to place + place + place. This item is determined (step S207). However, the configuration may be such that the control device 105 finally determines the character string as “place name” by referring to the arrangement table 107-6 in step S207.

  In the present embodiment, as shown in FIGS. 5 and 10, only the case where N = 6 has been described. Of course, even if the number N of partial character strings is other than 6, character string item determination is performed. Is possible. However, as N increases, the amount of processing increases due to an increase in the number of combinations. If the number of characters is large, it may be limited to a small number of characters, such as narrowing down to the first few characters.

[Embodiment 3]
Next, a third embodiment of the item determination device will be described. Also in the present embodiment, a mobile phone 100 will be described as an example of an item determination device. Mobile phone 100 in the present embodiment is different from mobile phone 100 in the first embodiment only in a part of the functional configuration and the processing procedure in the item determination process. That is, the overall configuration and hardware configuration of mobile phone 100 in the present embodiment is the same as mobile phone 100 in Embodiment 1 shown in FIGS. 1 and 2, and therefore description thereof will not be repeated here.

  The item determination device and the item determination processing according to the present embodiment are also realized by hardware such as the mobile phone 100 with a camera and software such as a control program. Generally, such software is stored in the second memory 107 at the time of manufacture or distributed by a user downloading it via a network or the like. Then, the software is read from the second memory 107 to the first memory 106 and executed by the control device 105.

<Functional configuration>
FIG. 16 is a block diagram showing a functional configuration of mobile phone 100 according to Embodiment 3. As illustrated in FIG. 16, the mobile phone 100 includes a photographing unit 142-1, an extraction unit 105-2, a recognition unit 105-3, a word search unit 105-8, a calculation unit 105-4, Determination unit 105-5, third determination unit 105-9, output unit 105-7, image storage unit 107-2, character recognition dictionary 107-3, result storage unit 107-4, trust A degree table 107-5, an arrangement table 107-6, and a word dictionary 107-7.

  Here, the extraction unit 105-2, the recognition unit 105-3, the word search unit 105-8, the calculation unit 105-4, the first determination unit 105-5, and the third determination unit 105-9. The output unit 105-7 refers to, for example, a control device 105 such as a CPU that reads a control program (software) stored in the second memory 107 into the first memory 106, and the control program (software). It is realized by executing. The image storage unit 107-2, the character recognition dictionary 107-3, the result storage unit 107-4, the reliability table 107-5, the arrangement table 107-6, and the word dictionary 107-7 include the first This is a function included in the storage unit 107-1 realized by the memory 106 or the second memory 107.

  Hereinafter, various functions of the mobile phone 100 will be described. As described above, the functions of the following blocks and the processing of each step are performed by a control program (software) stored in the second memory 107 by the control device 105. May be realized by reading out to the first memory 106 and executing the control program (software), or each may be realized by a dedicated hardware circuit or the like.

  It should be noted that the imaging unit 142-1, the extraction unit 105-2, the image storage unit 107-2, the character recognition dictionary 107-3, the result storage unit 107-4, and the reliability table 107- in the present embodiment. 5 and the arrangement table 107-6 have the same configuration as that of the first embodiment, and therefore description thereof will not be repeated here. Also, the method of generating the reliability table 107-5 in the present embodiment has the same configuration as that in the first embodiment, and therefore description thereof will not be repeated here.

  First, the word dictionary 107-7 is composed of at least one character, stores a plurality of words belonging to the first item, and holds a word as a character code string. The word dictionary 107-7 refers to a dictionary in which words belonging to at least one of the last name dictionary, first name dictionary, place name dictionary, general word dictionary, and the like are stored. The storage unit 107-1 of the mobile phone 100 only needs to store one of these dictionaries, and the words included in the dictionary are classified into the first item corresponding to the dictionary. It is understood that the word is suitable for being done. In addition, the word dictionary 107-7 may be adjusted such that the word dictionary 107-7 is limited to words that are highly likely to be written on the business card, for example, when the business card is to be read.

  The recognition unit 105-3 recognizes at least one first partial character string included in the character string and a character included in the first partial character string. Then, the recognition unit 105-3 acquires the position of each character included in the character string, and divides the character string into at least one first partial character string based on the position of each character. More specifically, the recognition unit 105-3 generates a character string as character code information by referring to the character recognition dictionary 107-3 with respect to the character string as image information extracted by the extraction unit 105-2. To do. Here, the character string recognition result by the recognition unit 105-3 includes character code information (may include recognition candidates) of each character determined by the recognition unit 105-3, coordinate information for each character, and the like. is there.

  And the recognition part 105-3 isolate | separates a character string into some 1st partial character strings from the character string arbitrary result memorize | stored in the result memory | storage part 107-4. As for the separation method for separating character strings, for example, after calculating the average distance from the distance between each character, it is larger than the average distance (the distance between characters of interest is equal to or more than a predetermined threshold). In some cases, there is a method of dividing a character string at that position. The character string separation result by the recognition unit 105-3 may be recorded in any form, but in the present embodiment, the number of separation positions as in the result storage unit 107-4 shown in FIG. The character string separation result is recorded by storing each separation position and the like.

  The word search unit 105-8 searches the text character string stored in the result storage unit 107-4 for a word stored in the word dictionary 107-7 in the text character string. Then, the search result is returned to the result storage unit 107-4. That is, the word search unit 105-8 is a determination unit that determines whether or not the first partial character string includes a word stored in the word dictionary.

  The calculation unit 105-4 refers to the reliability table 107-5 and determines at least one of the first partial character strings based on the reliability of each character included in the first partial character string for each first item. One string reliability is calculated and stored in the first memory 106 or the like. When the first partial character string includes a word stored in the word dictionary 107-7, the calculation unit 105-1 according to the present embodiment refers to the reliability table 107-5 and first Based on the character reliability corresponding to each character included in the second partial character string composed of characters excluding the characters included in the word stored in the word dictionary among the partial character strings, the second part The character string reliability for each first item of the character string is calculated. Then, the calculation unit 105-4 according to the present embodiment refers to the reliability table 107-5 when the first partial character string does not include a word stored in the word dictionary 107-7. At least one character string reliability of the first partial character string is calculated based on the reliability of each character included in the first partial character string for each first item.

  The first determination unit 105-5 determines the first item of the first partial character string based on the character string reliability. More specifically, the first determination unit 105-5 according to the present embodiment is based on the character string reliability when the first partial character string includes a word stored in the word dictionary 107-7. If the first item of the second partial character string is determined and the first partial character string does not include a word stored in the word dictionary 107-7, the first item is determined based on the character string reliability. The first item of the partial character string is determined.

  The output unit 105-7 is realized not only by the control device 105 but also by the monitor 150, a peripheral I / F circuit (not shown), and the like, and various kinds of information such as a recognition result stored in the result storage unit 122 are determined. The first item and the second item are transferred from the item determination program to another application or output to the outside of the mobile phone 100.

  When the first partial character string includes a word stored in the word dictionary 107-7, the third determination unit 105-9 includes the first item to which the word of the word dictionary 107-7 belongs, The second item of the partial character string is determined based on the first item of the partial character string. When the first partial character string does not include a word stored in the word dictionary 107-7, the third determination unit 105-9 performs the first item for each first item of at least one first partial character string. The second item of the character string is determined based on the reliability of the character string and the arrangement method of the first item of the arrangement table 107-6.

<Item judgment processing>
Next, an item determination process according to the present embodiment will be described. FIG. 17 is a flowchart showing a processing procedure of item determination processing in mobile phone 100 according to Embodiment 3. FIG. 18 is an image diagram showing a business card 203 to be subjected to the item determination process. In the present embodiment, it is assumed that mobile phone 100 stores only a surname dictionary in which only words suitable for being classified as surnames are stored as word dictionary 107-7. The input target image is assumed to be a business card image shown in FIG.

  In addition, as shown in FIG. 18, it is assumed that the business card 203 to be read has a stain on “(stock)”, and the item determination process when the character “(stock)” cannot be recognized will be described below. To do. In addition, since the font of “(stock)” is a special font that is not stored in the word dictionary, “(stock)” may not be correctly recognized and may be removed. In addition, since “(Co)” is a color character, it may not be recognized correctly. In addition, there are cases where it is not a recognition target as a part of the figure, such as a face photograph. As will be described below, the mobile phone according to the present embodiment can determine an item suitable for classifying a character string even when a part of the character string is unrecognizable.

  First, in step S101, the control device 105 extracts “building material manufacturing and sales”, “Uchiyamada building materials”, and “Kandagawa” from the character strings extracted by the photographing unit 142-1 and recognized by the control device 105. "Kenichi" is recognized as a partial character string. Then, the control device 105 stores the recognition results of the partial character strings of “building material manufacturing and sales”, “Uchiyamada building material”, and “Kandagawa Kenichi” realized by the first memory 106 or the second memory 107. Stored in the unit 107-1 (step S301).

  That is, in step S301, the control device 105 recognizes “building material manufacturing and sales” as a partial character string of n = 1 (first) and stores it in the storage unit 107-1. Further, the control device 105 recognizes “Uchiyamada Building Materials” as the n = 2 (second) partial character string and stores it in the storage unit 107-1. Further, the control device 105 recognizes “Kendagawa Kenichi” as the n = 3 (third) partial character string and stores it in the storage unit 107-1. That is, in this case, the number of partial character strings stored in the storage unit 107-1 is three (N = 3).

  In this example, the processing is performed when the control device 105 separates the recognized character string into a plurality of partial character strings. However, the control device 105 can separate the recognized character string because the distance between characters is constant. If not, the following processing is performed assuming that the character string includes one partial character string (assuming that there is only one partial character string).

  In the case where the character string itself obtained by the control device 105 is handled as a partial character string, the extracted character string itself is handled as a partial character string.

  Next, the control device 105 initializes the counter n to 1 (step S302). If n ≦ N (YES in step S303), control device 105 determines the word (last name) stored in word dictionary 107-6 (last name dictionary) for the nth partial character string. Whether or not is included (step S304). Here, it is assumed that two character strings “Uchiyamada” and “Kandagawa” are stored in the word dictionary 107-7, and the words stored in the word dictionary 107-7 in “building material manufacturing and sales”. (Last name) is not found.

  When the nth partial character string includes a word included in the word dictionary (YES in step S305), the character string reliability of the second partial character string including characters other than the word is calculated. (Steps S306 to S307). That is, the control device 105 obtains the character string reliability of the second partial character string for each item (step S306). The processing for obtaining the character string reliability of the second partial character string is, for example, for each item, adding the reliability of each character in the reliability table defined for each character and each item as shown in FIG. Do it. Then, the control device 105 determines that the first item having the maximum character string reliability is the first item suitable for classification of the second partial character string (step S307).

  If the nth partial character string does not include a word included in the word dictionary (NO in step S305), the partial character string itself is regarded as the first partial character string, and the reliability Is calculated (steps S308 to S309). That is, the control device 105 obtains the character string reliability of the first partial character string for each item (step S308). Then, the control device 105 determines that the first item having the maximum character string reliability is the first item suitable for classification of the first partial character string (step S309).

  Specifically, since the partial character string “building material manufacturing and sales” of n = 1 (first) does not include a word whose partial character string is included in the word dictionary 107-7 (step S305). In the case of NO), the character string reliability is calculated using the partial character string itself as the first partial character string (steps S308 to S309). That is, the control device 105 obtains the character string reliability of “building material manufacturing and sales” for each item (step S308). Then, the control device 105 determines that the first item having the maximum character string reliability is the first item suitable for being classified as “building material manufacturing and sales” (step S309).

  Since n = 2 (second) partial character string “Uchiyamada Building Materials” includes the word contained in the word dictionary 107-7 in the partial character string (in the case of YES in step S305). The character string reliability is calculated using “building material” composed of characters other than the word as the second partial character string (steps S306 to S307). That is, the control device 105 obtains the character string reliability of “building material” for each item (step S306). Then, the control device 105 determines the first item having the maximum character string reliability as the first item suitable for the classification of the “building material” as the second partial character string (step). S307).

  For n = 3 (third) partial character string “Kendagawa Kenichi”, since the partial character string includes a word included in the word dictionary 107-7 (YES in step S305). ), The character string reliability is calculated using “Kenichi” consisting of characters other than the word as the second partial character string (steps S306 to S307). That is, the control device 105 obtains the character string reliability of “Kenichi” for each item (step S306). Then, the control device 105 determines that the first item having the maximum character string reliability is the first item suitable for classifying “Kenichi” as the second partial character string ( Step S307).

FIG. 19 is an image diagram showing the reliability of characters constituting the partial character strings of n = 2 (second) and n = 3 (third). As shown in FIG.
Surname 5
Name 9
Place name 7
Word 10
Thus, it is determined that the second partial character string should be classified into the item of general word.

And for “Kenichi”,
Last name 10
Name 16
Place name 9
Word 11
Therefore, it is determined that the second partial character string should be classified into the item “name”.

  As described above, when the character string reliability of each of the first partial character string and the second partial character string and the respective first items are calculated, the control device 105 determines the character string reliability and the first character string reliability. 1 is stored in the storage unit 107-1, and the counter is incremented (n = n + 1) (step S310). However, as will be described later, here, the control device 105 calculates only the character string reliability of the first partial character string and the second partial character string, and stores the character string reliability in the storage unit 107-1. It may be configured to.

  On the other hand, if n> N (NO in step S303), control device 105 determines the first partial character string item, the second partial character string item, and the word determined in steps S307 and S309. The second item of the entire character string is determined from the combination (arrangement) with the items defined in the dictionary 107-7 (step S311). This is because, by using the arrangement table 107-6 as shown in FIG. 8, the control device 105 determines which item the partial character string including at least one first partial character string or second partial character string is assigned to. It is determined whether it should be classified (step S311).

  For example, in the case of “Uchiyamada building material”, which is a partial character string of n = 2 (second), the first item “last name” corresponding to “Uchiyamada” defined in the word dictionary 107-7. Next, since the first item “word” corresponding to “building material” is arranged, the control device 105 sets the n = 2 (second) partial character string to the second item “company name”. Judge that it should be classified. In the case of n = 3 (third) “Kendagawa Kenichi”, next to the first item “surname” corresponding to “Kandagawa” defined in the word dictionary 107-7, “ Since the first item “First Name” corresponding to “Kenichi” is arranged, the control device 105 should classify the n = 3 (third) partial character string into the second item “First Name First Name”. It is determined that

Further, in step S311, the control device 105 should be classified into the item "first name and last name" of the two partial character strings "Uchiyamada Kenki" and "Kendagawa Kenichi" that have been hit in the surname dictionary. It is good also as a structure which determines whether there exists. That is, since the first item that may be combined with the first item “last name” is “first name”, the control device 105 sets the second character string that forms the partial character string of n = 2 (second). A second partial character string having a higher reliability of the character string of “name” among the partial character string of 2 and the second partial character string constituting the n = 3 (third) partial character string The first item of the partial character string to be included may be determined as “first and last name”. That is, the character string reliability in the item of the names “building material” and “kenichi” which are the second partial character strings is:
Building materials ... 9
Kenichi ... 16
Therefore, “Kendagawa Kenichi” is more reliable than “Uchiyamada Construction Materials” as a first and last name, and the item “Kandagawa Kenichi” may be determined to be “first and last name”. .

  In the present embodiment, only the case where N = 3 has been described, but of course, it is possible even when N is 2 or less or when N is 4 or more. However, as N increases, the amount of processing increases due to an increase in the number of combinations. If the number of characters is large, it may be limited to a small number of characters, such as narrowing down to the first few characters.

  In the present embodiment, n = 2 (second) and n = 3 (third) partial character strings are derived from one first partial character string or one second partial character string. Although the case where it is configured has been described, even if the nth partial character string is further composed of a plurality of first partial character strings and a plurality of second partial character strings, item determination Can be done.

  Even if a conventional business card reader has a surname dictionary, it cannot determine whether “Uchiyamada Kenshi” or “Kandagawa Kenichi” is a surname without a name dictionary. . On the other hand, in the present embodiment, it is possible to determine which one of “Uchiyamada Construction Materials” and “Kandagawa Kenichi” is a surname without having a name dictionary. On the contrary, the name dictionary is used as usual, but the present embodiment can be partially applied such as using the configuration according to the present embodiment instead of using the surname dictionary. In addition, the mobile phone 100 according to Embodiment 1 and Embodiment 2 may be configured to partially use a word dictionary as in this embodiment.

<Summary of Embodiments 1 to 3>
Even if there are only a few hundred types of characters used, the number of types of words combined can reach tens of thousands or hundreds of thousands. Also, for example, if the location of Izumiotsu city in Izumiotsu city Minamisone 999 cannot be recognized due to dirt or the like, as in the business card shown in FIG. 10, two character strings “Minamisone” and “999” are displayed. It is necessary to determine which item the character string is based on. However, since the address 999 does not have keywords such as chome and address, it cannot be determined whether or not it is a “location name” based on the keyword.

  In the item determination apparatus according to the first to third embodiments, since the reliability of each character is stored for each item, the item determination is performed only by storing the character reliability × the number of items. It can be performed. In addition, the accuracy of item determination can be improved only by storing the arrangement of several tens of first items. That is, the item determination apparatus according to Embodiments 1 to 3 can perform partial character string or character string item determination even with a small storage capacity.

  Specifically, as described above, when a word dictionary is created, if a dictionary is created by a normal method, each of a surname dictionary, a name dictionary, a place name dictionary, a general word dictionary, etc., has a capacity of about several hundred Kbytes or more. is required. On the other hand, when the reliability of each character item is used, the numerical value of the reliability is, for example, a range of 1 byte (generally 0 to 255) or a range of 2 bytes (generally 0 to 65535). ), The storage capacity is about several Kbytes (ten and several Kbytes) for each dictionary. For example, even if there are 6000 characters to be recognized and four types of first items, the required storage capacity can be reduced to 1 × 6000 × 4 = 24 Kbytes. As described above, in the first to third embodiments, the dictionary (table) capacity required for the item determination can be greatly reduced as compared with the method using the conventional word dictionary.

  However, in the case of an unusual surname, the reliability of the characters that make up the surname may be low. Or, in the case of surnames (such as “Yamada”) that are often used in place names, such as “mountain”, “river”, “field”, etc., the place name is more reliable than the last name. There is a risk of end. In such a case, for example, by using only the surname dictionary and using this method instead of the name dictionary, the determination can be made without mounting the name dictionary. The surname dictionary is used in the same way as in the past, but by not using the surname dictionary, the capacity of the surname dictionary is the same as before, but the item determination device has the advantage that the storage capacity for the name dictionary can be reduced .

<Other embodiments>
In the above first to third embodiments, the item of the second or lower rank of the reliability of each character (or the character string reliability of each partial character string) is not used, but the character string of the partial character string (character string) It is also possible to use items below the second place for the calculation of reliability. At this time, when the first item of the second or lower reliability of each character is used rather than the first item of the reliability of each character, the character string reliability of the partial character string is higher. In addition, an item determination process may be performed in which the first item having the second or lower reliability is raised to the first place, and then the item of the partial character string is output.

  In the above embodiment, the application example of the item determination device to the folding mobile phone 100 has been described. However, the present invention is not limited to this, and can be similarly applied to a straight mobile phone and a slide mobile phone. Needless to say, the present invention can be applied not only to mobile phones but also to all electronic devices having a mail transmission function such as PDAs and notebook PCs. Furthermore, in the above-described embodiment, the present invention is applied to a program. However, it is of course possible to implement with a hard logic, and the program of the present invention is stored in an external memory such as a CD-ROM or an SD card. It is also possible.

  It goes without saying that the present invention can also be applied to a case where it is achieved by supplying a program to a system or apparatus. Then, a storage medium storing a program represented by software for achieving the present invention is supplied to the system or apparatus, and the computer (or CPU or MPU) of the system or apparatus stores the program code stored in the storage medium It is possible to enjoy the effects of the present invention also by reading and executing.

  In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiment, and the storage medium storing the program code constitutes the present invention.

  As a storage medium for supplying the program code, for example, a flexible disk, hard disk, optical disk, magneto-optical disk, CD-ROM, CD-R, magnetic tape, nonvolatile memory card (IC memory card), ROM (mask) ROM, flash EEPROM, etc.) can be used.

  Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also an OS (operating system) running on the computer based on the instruction of the program code However, it is needless to say that a case where the function of the above-described embodiment is realized by performing part or all of the actual processing and the processing is included.

  Further, after the program code read from the storage medium is written to a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function expansion is performed based on the instruction of the program code. It goes without saying that the CPU or the like provided in the board or the function expansion unit performs part or all of the actual processing and the functions of the above-described embodiments are realized by the processing.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

It is an external appearance perspective view which shows the mobile phone with a camera which concerns on this Embodiment. It is a block diagram which shows the hardware constitutions of the mobile telephone as an example of the communication apparatus which concerns on this Embodiment. 3 is a block diagram illustrating a functional configuration of the mobile phone according to Embodiment 1. FIG. It is an image figure which shows the data structure of the character string recognition result database memorize | stored in a result memory | storage part. It is an image figure which shows the character string and partial character string in image data. It is an image figure which shows the data structure of a reliability table. It is an image figure which shows an example of the database of a Japanese surname. It is an image figure which shows the data structure of an arrangement | sequence table. 3 is a flowchart showing a processing procedure of item determination processing in the mobile phone according to Embodiment 1; FIG. 3 is an image diagram showing a business card that is an object of an item determination process according to the first embodiment. It is an image figure which showed the reliability of the character which comprises the 1st partial character string which concerns on Embodiment 1, and the character string reliability of the 1st partial character string. It is an image figure which showed the reliability of the character which comprises the 2nd partial character string which concerns on Embodiment 1, and the character string reliability of the 2nd partial character string. 6 is a block diagram illustrating a functional configuration of a mobile phone according to Embodiment 2. FIG. 6 is a flowchart showing a processing procedure of item determination processing in the mobile phone according to the second embodiment. It is an image figure which showed the reliability of the character which comprises the partial character string which concerns on Embodiment 2. FIG. 7 is a block diagram showing a functional configuration of a mobile phone according to Embodiment 3. FIG. 12 is a flowchart showing a processing procedure of item determination processing in the mobile phone according to the third embodiment. FIG. 10 is an image diagram showing a business card that is an object of an item determination process according to the third embodiment. It is the image figure which showed the reliability of the character which comprises the partial character string of n = 2 (2nd) and n = 3 (3rd) based on Embodiment 3. FIG.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Antenna, 100 Mobile phone, 101 Wireless communication part, 105 Control apparatus, 105-2 Extraction part, 105-3 Recognition part, 105-4 Calculation part, 105-5 1st determination part, 105-6 2nd determination Section, 105-7 output section, 105-8 word search section, 105-9 third determination section, 106 first memory, 107 second memory, 107-1 storage section, 107-2 image storage section, 107 -3 character recognition dictionary, 107-4 result storage unit, 107-5 reliability table, 107-6 arrangement table, 107-7 word dictionary, 114 first case, 116 second case, 120 input unit, 142 1st camera, 142-1 imaging | photography part, 143 2nd camera, 150 monitor, 172 microphone, 173 1st speaker, 174 2nd speaker.

Claims (11)

An item determination device for determining an item of a character string,
Recognizing means for recognizing each character included in the first partial character string for each of at least one first partial character string included in the input character string;
A reliability table in which the reliability of each character, which is determined in advance based on a first item in which at least one word including each character is to be classified, is stored for each first item;
Referring to the reliability table, at least one character string reliability of the first partial character string is determined based on the reliability of each character included in the first partial character string for each first item. A calculating means for calculating;
An item determination apparatus comprising: first determination means for determining an item of the first partial character string based on the character string reliability.   The item determination device according to claim 1, wherein the reliability of the first item of each character is determined based on a type of a word including the character and a degree to which the word is used. The recognition unit obtains the position of each character included in the character string, dividing the said string to at least one of said first sub-string on the basis of the position in the character string of each character,
The calculation means refers to the reliability table, calculates a character string reliability based on the reliability of each character included in the first partial character string for each of the first items,
The said 1st determination means determines the 1st item of the said 1st partial character string based on the said character string reliability for every said 1st item, The Claim 1 or Claim 2 characterized by the above-mentioned. Item determination device. A first arrangement table that stores second items associated with arrangements of the plurality of first items;
The second determination means for determining the second item of the character string based on the character string reliability for each first item of the first partial character string and the arrangement of the first items; The item determination device according to claim 3 provided. A first arrangement table that stores second items associated with the arrangement of the plurality of first items;
The calculation means, for each arrangement of the first items stored in the first arrangement table, based on the arrangement of the first items stored in the first arrangement table. Calculate the string reliability of 1 substring,
The first determination means determines the second item of the first partial character string based on the character string reliability for each arrangement of the first items stored in the first arrangement table. The item determination apparatus according to claim 1, wherein the determination is performed. A word dictionary storing a plurality of words composed of at least one character belonging to any of the first items;
Determining means for determining whether or not the first partial character string includes a word stored in the word dictionary;
The calculating means includes
When the first partial character string includes a word stored in the word dictionary, referring to the reliability table, the word stored in the word dictionary in the first partial character string Based on the reliability corresponding to each character included in the second partial character string composed of characters excluding the included characters, the character string reliability for each first item of the second partial character string is determined. Calculate
When the first partial character string does not include a word stored in the word dictionary, referring to the reliability table, the first item of each character included in the first partial character string And calculating at least one character string reliability of the first partial character string based on the reliability of
The first determination means includes
When the first partial character string includes a word stored in the word dictionary, the first item of the second partial character string is determined based on the character string reliability;
The first item of the first partial character string is determined based on the character string reliability when the first partial character string does not include a word stored in the word dictionary. Or the item determination apparatus of Claim 2. A first arrangement table that stores second items associated with arrangements of the plurality of first items;
When the first partial character string includes a word stored in the word dictionary, based on the first item to which the word of the word dictionary belongs and the first item of the second partial character string The item determination apparatus according to claim 6, further comprising: a third determination unit that determines a second item of the partial character string.   8. The item determination device according to claim 1, wherein each of the first items is one of a last name, a first name, a place name, a prefecture name, a municipality name, and a general word. An item determination program for causing a computer to determine an item of a character string,
The computer
A reliability table in which the reliability of each character determined in advance based on a first item in which at least one word including each character is to be classified is stored for each first item;
The item determination program is stored in the computer.
Recognizing each character included in the first partial character string for each of at least one first partial character string included in the input character string;
Referring to the reliability table, at least one character string reliability of the first partial character string is determined based on the reliability of each character included in the first partial character string for each first item. A calculating step;
And a step of determining an item of the first partial character string based on the character string reliability. A computer-readable recording medium recording an item determination program for causing a computer to determine an item of a character string,
The computer
A reliability table in which the reliability of each character determined in advance based on a first item in which at least one word including each character is to be classified is stored for each first item;
The item determination program is stored in the computer.
Recognizing each character included in the first partial character string for each of at least one first partial character string included in the input character string;
Referring to the reliability table, at least one character string reliability of the first partial character string is determined based on the reliability of each character included in the first partial character string for each first item. A calculating step;
And a step of determining an item of the first partial character string based on the character string reliability. An item determination method using an item determination device for determining an item of a character string,
Item judgment device
A reliability table in which the reliability of each character, which is determined in advance based on a first item in which at least one word including each character is to be classified, is stored for each first item;
A control device,
Recognizing each character included in the first partial character string for each of the at least one first partial character string included in the input character string;
The control device refers to the reliability table and determines at least one of the first partial character strings based on the reliability of each character included in the first partial character string according to the first item. Calculating string reliability; and
The control device includes a step of determining an item of the first partial character string based on the character string reliability.

Images