JP5101584B2 - Dictionary update device, dictionary update program, character recognition device, and character recognition program - Google Patents

Description

  The present invention relates to a dictionary update device that updates a dictionary used in character recognition processing and the like, and a character recognition device having this dictionary update function.

  Patent Document 1 discloses a handwritten information recognition device. This device is connected to a terminal by communication. The apparatus includes an information communication unit that communicates with a caller, an information editing unit that edits information received by the information communication unit, a handwritten information recognition unit that recognizes a handwritten character pattern, Attribute addition means for adding attributes and recognition that classifies handwritten information according to the attributes of the caller and recognizes the classified handwritten information using a recognition dictionary file specialized for each caller's handwritten character trap Control means. Since the handwritten information recognition apparatus provided with such means can be recognized using a recognition dictionary file specialized for a caller's handwritten character, accurate recognition is possible.

  However, even if a recognition dictionary file specialized for a caller's handwritten character trap is used, terms that are not registered in the file cannot be recognized. Normally, generic terms are registered in the recognition dictionary file, but there are many cases where special terms that are valid only for a specific industry are not registered.

  For this reason, for example, characters handwritten on the whiteboard in a meeting may be digitized by automatic recognition technology after the meeting and used to create minutes. However, since special terms are often used in a meeting, character recognition There is a high probability of failure. Such a problem may be obtained by appropriately updating the recognition dictionary file. However, the updating of the dictionary file is very complicated and costly, so the fact is that it has not been updated.

  The present invention has been made based on such circumstances, and an object of the present invention is to make it possible to easily update a recognition dictionary file at a low cost and to improve character recognition accuracy of handwritten characters and the like.

The present invention provides an input means for extracting accompanying data accompanying the image data from the input image data, and metadata generation for generating metadata in which attributes and values of the accompanying data are set from the accompanying data. Obtained from the external system by the external system cooperation means for giving the command defined in the dictionary selection rule to the external system defined in the dictionary selection rule and obtaining the result The dictionary selection rule is determined based on the information and the metadata generated by the metadata generation unit, and the dictionary selection unit that selects a dictionary file defined in the rule to be established, and the dictionary selection unit selects the dictionary file. The dictionary update rule set for the dictionary file is determined, and the update is defined according to the updated dictionary update rule. Is obtained by a dictionary updating means for updating the set dictionary file in the rule.
The present invention also provides an image acquisition means for acquiring image data input via the input means, and the image data obtained by the image acquisition means using the dictionary file updated by the dictionary update means. A character recognition means for recognizing,

  According to the present invention in which such measures are taken, the recognition dictionary file can be easily updated at low cost, and the accuracy of recognizing characters such as handwritten characters can be improved.

1 is a system configuration diagram of a first embodiment according to the present invention. FIG. The block diagram which shows the component required in order to implement | achieve a dictionary update function in the server of the embodiment. The figure which shows an example of the dictionary file for recognition preserve | saved in the dictionary database in the same embodiment. The figure which shows an example of the rule data recorded on the dictionary selection rule file in the same embodiment. The figure which shows an example of the rule data recorded on the dictionary connection rule file in the same embodiment. The figure which shows an example of the rule data recorded on the dictionary update rule file in the same embodiment. The figure which shows an example of the cooperation information data file preserve | saved in the cooperation information database in the embodiment. The flowchart which shows the procedure of the basic process regarding the dictionary update function of a server in the embodiment. 5 is a flowchart showing in detail a procedure of dictionary selection processing in basic processing in the embodiment. In the same embodiment, the flowchart which shows the procedure of the rule connection process in a basic process in detail. 5 is a flowchart showing in detail a procedure of a selection dictionary candidate acquisition process in the rule connection process in the embodiment. 5 is a flowchart showing in detail a procedure of dictionary update processing in basic processing in the embodiment. The system block diagram of 2nd Embodiment which concerns on this invention. The block diagram which shows the component required in order to implement | achieve a dictionary update function and a character recognition function in the server of the embodiment. 5 is a flowchart showing a procedure of basic processing related to a dictionary update function and a character recognition function of the server in the embodiment.

Hereinafter, embodiments according to the present invention will be described with reference to the drawings.
(First embodiment)
First, a first embodiment according to the dictionary updating apparatus of the present invention will be described with reference to FIGS.

  FIG. 1 is a system configuration diagram of the first embodiment. This system includes a server 1, a client 2, and a multifunction device 3, which are connected by a network 4. The network 4 may be an intranet or the Internet.

  The server 1 can exchange electronic mail with the client 2 using SMTP (Simple Mail Transfer Protocol). Further, the server 1 can take in the original reading data read by the scanner function provided in the multifunction machine 3.

  The number of clients 2 and multifunction devices 3 is not limited to one, and a plurality of clients 2 and multifunction devices 3 may be connected to the server 1 via the network 4. In this case, the network that connects the server 1 and each client 2 and the network that connects the server 1 and each multifunction device 3 may be provided in different systems.

  The server 1 functions as a dictionary update device. That is, the server 1 uses one or more external dictionary files including a groupware system, a Web dictionary system, and the like based on data received from the client 2 and the multifunction device 3 for various recognition dictionary files registered in the dictionary database 5. It has a dictionary update function 7 for updating using the system 6 as appropriate. The dictionary update function 7 is realized by a dictionary update program installed in the server 1.

  The dictionary update program may be stored in advance in a storage device inside the server 1 or may be downloaded from the network to the server 1. Further, what is stored in the recording medium may be installed in the server 1. The recording medium may be in any form as long as it can store a program such as a CD-ROM and can be read by the server 1.

  FIG. 2 is a block diagram showing components necessary for realizing the dictionary update function 7 in the server 1. The server 1 includes an input unit 11, a metadata generation unit 12, a dictionary selection unit 13, a dictionary update unit 14, a related information search unit 15, an external system cooperation unit 16 for each external system 6, and a database storage unit 17. .

  The database storage unit 17 stores a rule database 20 and a linkage information database 30 in addition to the dictionary database 5. In FIG. 1, the database storage unit 17 is provided in the storage device in the server 1, but the database storage unit 17 may be provided in an auxiliary storage device externally attached to the server 1. Alternatively, for example, the rule database 20 and the linkage information database 30 are stored in a storage device in the server 1, and the dictionary database 5 is stored in an auxiliary storage device outside the server 1. It can also be provided outside.

  An example of a recognition dictionary file stored in the dictionary database 5 is shown in FIG. The dictionary file includes a dictionary name file 71, a word file 72, a dictionary / word related file 73, and the like. In the dictionary name file 71, a dictionary type and a dictionary name are recorded in association with the dictionary ID. In the word file 72, a word or the like is recorded in association with the word ID. In the dictionary / word related file 73, the appearance frequency and the like are recorded in association with the dictionary ID and the word ID.

  The dictionary ID is a code for identifying a dictionary having a corresponding dictionary type and dictionary name. The word ID is a code for identifying the corresponding word. The appearance frequency is the frequency at which the word specified by the corresponding word ID appears in the dictionary specified by the corresponding dictionary ID.

  The rule files stored in the rule database 20 include a dictionary selection rule file 21, a dictionary connection rule file 22, and a dictionary update rule file 23. An example of rule data recorded in the dictionary selection rule file 21 is shown in FIG. As shown in the figure, the dictionary selection rule file 21 records at least a target data code, an attribute, a comparison value, a condition, an external linkage ID, and a selection dictionary ID in a one-to-one correspondence with a unique rule ID. .

  The target data code is a code that specifies data to be compared. The code “0” indicates that the comparison target data is metadata. The code “1” is obtained from the external system. The code “2” indicates that the comparison target data is both metadata and data acquired from the external system. The attribute is information that specifies an attribute to be compared.

  The comparison value is information for specifying a value to be compared with the comparison target data. For example, a character string such as “Taro.tokyo@ttt.co.jp” or “Tokyo Taro” indicating an email address, or “2 Information such as numerical values such as “100” is set. If no comparison value is specified, data “NULL” is set. The condition is information for specifying the relationship in which the attribute value of the comparison target data and the comparison value are established. For example, information indicating a condition for string comparison such as “complete match” or “partial match” is set. , Information indicating a numerical comparison condition such as “difference is within comparison value” or “difference is greater than or equal to comparison value” is set. Note that data “Not” is set when the condition is not questioned because the comparison value is not designated.

  The external cooperation ID is a code for identifying the external system 6 that is the cooperation destination in the case of cooperation with the external system 6. Each external system 6 is assigned with a unique external linkage ID. If the external system 6 is not linked, “0” is set as the external linkage ID. The selected dictionary ID is a code for identifying a dictionary file to be selected when a condition is established between the comparison target data and the comparison value. Each dictionary file stored in the dictionary database 5 is assigned with a unique dictionary ID.

  For example, in the case of a dictionary selection rule corresponding to the rule ID “1”, the target data code is “0”, the attribute is “sender address”, the comparison value is “Taro.tokyo@ttt.co.jp”, and the condition is “ Condition is required when the sender address in the metadata is exactly the same as “Taro.tokyo@ttt.co.jp” because “exact match”, the external linkage ID is “0”, and the selected dictionary ID is “1”. Is established, and the dictionary file identified by the dictionary ID “1” is selected.

  In the case of the dictionary selection rule corresponding to the rule ID “2”, the target data code is “0”, the attribute is “destination address”, the comparison value is “Group1”, the condition is “forward match”, and the external linkage ID is “ Since the selected dictionary ID is “0”, the condition is satisfied when the destination address in the metadata starts with “Group1”, and the dictionary file identified by the dictionary ID “2” is selected.

  In the case of the dictionary selection rule corresponding to the rule ID “11”, the target data code is “2”, the attribute is “Date”, the comparison value is “2”, the condition is “difference is within the comparison value”, and the external linkage ID Is “1” and the selected dictionary ID is “1”, the difference between the date in the metadata and the date of the data acquired from the external system 6 identified by the external linkage ID “1” is within 2 days. In some cases, the condition is satisfied, and the dictionary file identified by the dictionary ID “1” is selected.

  Similarly, in the case of the dictionary selection rule corresponding to the rule ID “13”, the target data code is “1”, the attribute is “group name”, the comparison value is “Null”, the condition is “Not”, and the external linkage ID is “ Since the selected dictionary ID is “3”, the dictionary file identified by the dictionary ID “3” is selected for the group name of the data acquired from the external system 6 identified by the external cooperation ID “2”. Is done.

  An example of rule data recorded in the dictionary connection rule file 22 is shown in FIG. As shown in the figure, the dictionary connection rule file 22 has a one-to-one correspondence with the unique connection rule ID, each rule ID (rule IDa, rule IDb) of two dictionary selection rules to be connected, and a condition. And a dictionary selection method are recorded.

  The condition is information indicating a relationship in which two dictionary selection rules to be connected are established, information “OR” that the connection condition is deemed to be satisfied when at least one dictionary selection rule is established, and both dictionary selections There is “AND” which considers that the connection condition is satisfied when the rule is satisfied. The dictionary selection method is information for designating a dictionary file to be selected when the connection condition is satisfied, and includes “Both”, “TrueOnly”, “Rule2Only”, and the like. “Both” selects two dictionary files respectively selected by the two dictionary selection rules to be connected when the connection condition is satisfied. “TrueOnly” selects a dictionary file corresponding to the rule that is established out of the two dictionary selection rules to be connected when the connection condition is satisfied. “Rule2Only” selects a dictionary file corresponding to the rule IDb among the two dictionary selection rules to be connected when the connection condition is satisfied.

  For example, in the case of the dictionary concatenation rule corresponding to the concatenation rule ID “101”, the rule ID “1”, the rule ID b is “2”, the condition “OR”, and the dictionary selection method “Both”. When at least one of the dictionary selection rule of “” and the dictionary selection rule of rule ID “2” is established, the selection dictionary ID of each of the dictionary selection rule of rule ID “1” and the dictionary selection rule of rule ID “2” Select the dictionary file identified by.

  Further, in the case of the dictionary concatenation rule corresponding to the concatenation rule ID “102”, the rule ID “11”, the rule ID b is “12”, the condition “AND”, and the dictionary selection method “Both”. Only when the dictionary selection rule of “” and the dictionary selection rule of rule ID “12” are both established, the selection dictionary ID of the dictionary selection rule of rule ID “11” and the dictionary selection rule of rule ID “12” Select a dictionary file to be identified.

  Similarly, in the case of the dictionary concatenation rule corresponding to the concatenation rule ID “103”, the rule ID “3”, the rule ID b is “4”, the condition “OR”, and the dictionary selection method “TrueOnly”. When at least one of the 3 ”dictionary selection rule and the rule ID“ 4 ”dictionary selection rule is established, the dictionary file identified by the selected dictionary ID of the established dictionary selection rule is selected.

  Further, in the case of the dictionary concatenation rule corresponding to the concatenation rule ID “104”, the rule ID “13”, the rule ID b is “14”, the condition “AND”, and the dictionary selection method “Rule 2 Only”. The dictionary file identified by the selected dictionary ID of the dictionary selection rule of rule ID “14” is selected only when both the dictionary selection rule of “” and the dictionary selection rule of rule ID “14” are satisfied.

  An example of rule data recorded in the dictionary update rule file 23 is shown in FIG. In the dictionary update rule file 23, at least a target data code, an attribute, a comparison value, a condition, an update content, and an external cooperation ID are recorded in one-to-one correspondence with a unique dictionary ID. The target data code, attribute, comparison value, condition, and external linkage ID are the same as the data recorded in the dictionary selection rule file 21. The update contents are information indicating contents to be updated when the condition is satisfied. For example, “additional update” adds an attribute word in the target data to the dictionary file specified by the dictionary ID. Alternatively, if an attribute word has already been registered in the dictionary file, the subsequent frequency is incremented by +1.

  For example, in the case of the first dictionary update rule in FIG. 6, the dictionary ID is “1”, the target data code is “0”, the attribute is “frequent word”, the comparison value is “Null”, the condition is “Not”, and the update is performed. Since the contents are “additional update” and the external linkage ID is “0”, when the dictionary file with the dictionary ID “1” is the update target, the frequent word of the metadata is additionally updated in the dictionary file.

  In the case of the second dictionary update rule from the top in FIG. 6, the dictionary ID is “3”, the target data code is “1”, the attribute is “person name”, the comparison value is “Null”, and the condition is “Not”. Since the update content is “additional update” and the external linkage ID is “3”, when the dictionary file with the dictionary ID “3” is an update target, the process returns from the external system 6 identified by the external linkage ID “3”. The person name acquired as a value is added to and updated in the dictionary file.

  Similarly, in the case of the third dictionary update rule from the top in FIG. 6, the dictionary ID is “1”, the target data code is “1”, the attribute is “synonym”, the comparison value is “Null”, and the condition is “ “Not”, the update content is “additional update”, and the external linkage ID is “1”. Therefore, when the dictionary file with the dictionary ID “1” is the update target, the external system 6 identified by the external linkage ID “1”. The synonym acquired as a return value from is added to the dictionary file and updated.

  In the case of the fourth dictionary update rule from the top in FIG. 6, the dictionary ID is “2”, the target data code is “1”, the attribute is “related word”, the comparison value is “Null”, and the condition is “Not”. Since the update content is “additional update” and the external linkage ID is “2”, when the dictionary file with the dictionary ID “2” is the update target, the external system 6 identified by the external linkage ID “2” The related word acquired as a return value is added to and updated in the dictionary file.

  An example of the cooperation information data file 31 stored in the cooperation information database 30 is shown in FIG. As shown in the figure, in the cooperation information data file 31, data including service destination, action, and attribute items is set and stored in a one-to-one correspondence with the external cooperation ID for identifying each external system 6. Yes. The item “service destination” is information indicating the form of the external system 6. For example, “GroupWare” indicates that the external system 6 identified by the corresponding external linkage ID is a groupware system, and “dictionary service”. Indicates that the external system 6 identified by the corresponding external linkage ID is a dictionary service system on the Web.

  Groupware is software for supporting information sharing and collaborative work among members in an organization such as a company. A groupware system is a system in which groupware is installed. The functions provided by the groupware include functions such as a web bulletin board, e-mail transfer, schedule management, and member management.

  The dictionary service system is a system that returns data such as the meaning of a specified word and synonyms when a query including a search word is transmitted to a dictionary site on the Web using a public API (Application Program Interface). As another external system, for example, there is a Web search service system. In this system, when a query including a search word is transmitted to a search site on the Web using a public API, the number of search results, information on a related Web page, or after a search, etc., as a search result of a specified word A system that returns data.

  The item “action” is information for specifying an action to be executed by the external system 6 identified by the corresponding external cooperation ID, and the item “attribute” is information on a data item applied in the action.

  For example, in the case of cooperation information data including the external cooperation ID “1”, the service destination “GroupWare1”, the action information “acquire schedule information”, and the attribute “transmission date / time” are set. This means that an action of acquiring schedule information is executed for the identified groupware system, and the transmission date and time in the schedule information is acquired as a return value from the groupware system.

  In the case of linkage information data including the external linkage ID “3”, the service destination “dictionary service”, the action information “check whether registered”, and the attribute “frequent word” are set. The dictionary service system on the web identified by the ID “3” checks whether or not a frequent word of metadata is registered, and if it is registered, the dictionary service system uses the word as a return value. Means to get.

  Next, functions of each component (input unit 11, metadata generation unit 12, dictionary selection unit 13, dictionary update unit 14, related information search unit 15, external system cooperation unit 16) other than the database storage unit 17 of the server 1 Will be described with reference to the flowcharts of FIGS. FIG. 8 is a flowchart showing the procedure of basic processing related to the dictionary update function 7 of the server 1, FIG. 9 is a flowchart showing details of the procedure of dictionary selection processing in the basic processing, and FIG. FIG. 11 is a flowchart showing in detail the procedure of the selected dictionary candidate acquisition process in the rule connection process, and FIG. 12 is a flowchart of the dictionary update process in the basic process. It is a flowchart which shows a procedure in detail.

  The input unit 11 waits for data related to dictionary update to be input (FIG. 8: ST1). Data relating to dictionary update is, for example, e-mail data sent from the client 2, document reading data sent from the scanner function of the multifunction machine 3, or the like.

  When data related to dictionary update is input via the network 4, the input unit 11 analyzes the input data (FIG. 8: ST2). Then, image data is extracted from the input data (FIG. 8: ST3). Further, the accompanying data is extracted from the input data (FIG. 8: ST4).

  The image data is data in a binary file format that can be analyzed by a computer. For e-mail data, an image file attached to the body of an e-mail is applicable, and for scanned document data, scanned document image data is Applicable. The accompanying data is data accompanying the input data, and for the email data, the sender email address, the recipient email address, the transmission date and time, the title, the email text, etc. included in the header information correspond to the document reading data. For example, the reading date and time, the document size, the number of read pages, the type of reading device, and the like included in the index data are applicable (input means).

  The input unit 11 transmits accompanying data extracted from the input data to the metadata generation unit 12. The metadata generation unit 12 generates metadata of the input data based on the accompanying data received from the input unit 11 (FIG. 8: ST5). In the metadata, the attribute of the accompanying data and its value are set. For example, when the attribute of the accompanying data is “sender mail address”, the value is a mail address described in “From” included in the header information of the electronic mail. In the case of the attribute “recipient mail address”, the mail address is described in “To” or “Cc” included in the header information. In the case of the attribute “transmission date / time”, the date is described in “Date” included in the header information. In the case of the attribute “title”, it is a character string described in “Subject” included in the header information.

  For the mail text, morphological analysis is performed, and the word, part of speech, and number of appearances are calculated. Then, the noun having the highest number of appearances with respect to the attribute “most frequently used word” is set as the value. In addition, a word having a certain number of appearances or more with respect to the attribute “frequent word” is set as the value. Further, when a person name is obtained from morphological analysis, the value obtained from the morphological analysis for the attribute “person name” is used as the value (metadata generating means).

  The metadata generation unit 12 transmits the generated metadata to the dictionary selection unit 13. The dictionary selection unit 13 selects one or more dictionary files to be updated based on the dictionary selection rules recorded in the dictionary selection rule file 21 (FIG. 8: ST6).

  Specifically, first, the selected dictionary memory is cleared (FIG. 9: ST11). Further, the number of data N of the dictionary selection rule is stored in the threshold memory (FIG. 9: ST12). Further, the count value i of the counter memory is reset to “0” (FIG. 9: ST13). The selection dictionary memory, threshold memory and counter memory are all formed in an internal RAM (Random Access Memory) of the server 1.

  When the processes of ST11 to ST13 are completed, the dictionary selecting unit 13 increments the count value i of the counter memory by “1” (FIG. 9: ST14). Then, each time counting up, it is compared with the value N in the threshold memory (FIG. 9: ST15), and the processing of ST16 to ST29 is repeated until the count value i exceeds the threshold value N.

  That is, the dictionary selection unit 13 reads i-th (i is a count value) dictionary selection rule data from the dictionary selection rule file 21 and stores it in the selection rule memory SERule-i (FIG. 9: ST16). Then, the external linkage ID of the dictionary selection rule data stored in the selection rule memory SERule-i is checked (FIG. 9: ST17).

  When the external linkage ID is “0” (NO in ST17), the dictionary selection rule data need not be linked with the external system 6. In this case, the dictionary selection unit 13 recognizes the attribute of the dictionary selection rule data stored in the selection rule memory SERule-i. Then, the value of this attribute is extracted from the metadata received from the metadata generation unit 12 and stored in the first comparison memory mValue (FIG. 9: ST18). Further, the comparison value of the dictionary selection rule data stored in the selection rule memory SERule-i is stored in the second comparison memory rValue (FIG. 9: ST19).

  On the other hand, when the external linkage ID of the dictionary selection rule data stored in the selection rule memory SERule-i is “1” or more (YES in ST17), the dictionary selection rule data is linked to the external system 6. In this case, the dictionary selection unit 13 transmits the metadata received from the metadata generation unit 12 and the external linkage ID of the dictionary selection rule data to the related information search unit 15 (FIG. 9: ST20). Then, it waits for a return value from the related information search unit 15 (ST21 in FIG. 9).

  When the related information search unit 15 receives the metadata and the external cooperation ID from the dictionary selection unit 13, the related information search unit 15 searches the cooperation information data file 31 of the cooperation information database 30 and stores it in correspondence with the received external cooperation ID. Get external linkage system information such as service destination, action, and attribute. Then, the external cooperation system information and the metadata received from the dictionary selection unit 13 are transmitted to the corresponding external system cooperation unit 16.

  When the external system cooperation unit 16 receives the external cooperation system information and metadata from the related information search unit 15, the external system cooperation unit 16 transmits the external cooperation system information to the external system 6 identified by the external cooperation ID of the external cooperation system information. Give action instructions. If a return value as a result of the action is acquired from the external system 6, the return value is given to the dictionary selection unit 13 via the related information search unit 15 (external system cooperation unit).

  When the dictionary selection unit 13 receives a return value from the external system 6 via the related information search unit 15, the dictionary selection unit 13 stores the return value in the return value memory sValue (FIG. 9: ST22). Next, the dictionary selection unit 13 determines whether the target data code of the dictionary selection rule data stored in the selection rule memory SERule-i is “1” or “2” (FIG. 9: ST23).

  When the target data code is “1” (YES in ST23), the dictionary selection unit 13 stores the value of the return value memory sValue in the first comparison memory mValue (FIG. 9: ST24). Further, the comparison value of the dictionary selection rule data stored in the selection rule memory SERule-i is stored in the second comparison memory rValue (FIG. 9: ST25).

  When the target data code is “2” (NO in ST23), the dictionary selection unit 13 acquires the attribute of the dictionary selection rule data stored in the selection rule memory SERule-i. Then, a value corresponding to the attribute is extracted from the metadata received from the metadata generation unit 12, and stored in the first comparison memory mValue (FIG. 9: ST26). Further, the value of the return value memory sValue is stored in the second comparison memory rValue (FIG. 9: ST27).

  When the values are stored in the first comparison memory mValue and the second comparison memory rValue, respectively, the dictionary selection unit 13 compares the value of the first comparison memory mValue with the value of the second comparison memory rValue. Then, it is determined whether or not the conditions of the dictionary selection rule data stored in the selection rule memory SERule-i are satisfied (FIG. 9: ST28). If the condition is not satisfied (NO in ST28), the process returns to ST14, further increments the counter i, and executes the processes after ST15 again.

  When the condition is satisfied (YES in ST28), the dictionary selection unit 13 adds the rule ID of the dictionary selection rule data stored in the selection rule memory SERule-i and the selection dictionary ID as a pair and adds them to the selection dictionary memory ( ST29). Thereafter, the process returns to ST14, the counter i is further counted up, and the processes after ST15 are executed again.

  Thus, when the processing of ST16 to ST29 is executed for all dictionary selection rule data recorded in the dictionary selection rule file 21, the counter i exceeds the threshold value N (YES in ST15), and the dictionary selection processing is performed. End (dictionary selection means).

  When the dictionary selection process ends, the dictionary selection unit 13 next executes a rule connection process (FIG. 8: ST7). First, the data number N of dictionary connection rules recorded in the dictionary connection rule file 22 is stored in the threshold memory (ST31 in FIG. 10). Further, the count value i of the counter memory is reset to “0” (FIG. 10: ST32).

  When the processes of ST31 to ST32 are completed, the dictionary selection unit 13 counts up the count value i of the counter memory by “1” (FIG. 10: ST33). Then, each time counting up, it is compared with the value N in the threshold memory (FIG. 10: ST34), and the processes of ST35 to ST38 are repeatedly executed until the count value i exceeds the threshold value N.

  That is, the dictionary selection unit 13 reads the i-th (i is a count value) connected dictionary rule data from the dictionary connection rule file 22 and stores it in the connection rule memory CORule-i (FIG. 10: ST35). Then, with respect to the linked dictionary rule data stored in the linked rule memory CORule-i memory, the linked rule ID, each ID of the two dictionary selection rules to be linked (rule IDa, rule IDb), each condition, and information on the dictionary selection method Are respectively stored in the ID memory ID, the first target memory Rule-ia, the second target memory Rule-ib, the condition memory Condition, and the dictionary selection method memory Dictionary (FIG. 10: ST36).

  Thereafter, the dictionary selection unit 13 executes a selection dictionary candidate acquisition process (FIG. 10: ST37). Specifically, first, the selected dictionary candidate memory is cleared (FIG. 11: ST51). Next, it is determined whether the condition information stored in the condition memory Condition is “AND” (FIG. 11: ST52) or “OR” (FIG. 11: ST53). If the condition information is neither “AND” nor “OR” (FIG. 11: NO in ST52, NO in ST53), the selected dictionary candidate acquisition process ends.

  When the condition information is “AND” (YES in ST52), the dictionary selecting unit 13 stores the rule ID stored in the first target memory Rule-ia and the second target memory Rule-ib in the selected dictionary memory. It is determined whether or not both rule IDs are present (FIG. 11: ST54). When at least one of the rule ID stored in the first target memory Rule-ia and the rule ID stored in the second target memory Rule-ib is not stored in the selected dictionary memory (NO in ST54), The selected dictionary candidate acquisition process ends.

  When both the rule ID stored in the first target memory Rule-ia and the rule ID stored in the second target memory Rule-ib are stored in the selected dictionary memory (YES in ST54), the dictionary The process proceeds to information determination processing (FIG. 11: ST56, ST57, ST58) indicating the dictionary selection method stored in the selection method memory Dictionary.

  On the other hand, when the condition is “OR” (YES in ST53), the dictionary selection unit 13 uses the rule ID stored in the first target memory Rule-ia and the second target memory Rule-ib in the selected dictionary memory. It is determined whether or not at least one of the stored rule IDs exists (FIG. 11: ST55). If both the rule ID stored in the first target memory Rule-ia and the rule ID stored in the second target memory Rule-ib are not stored in the selected dictionary memory (NO in ST55), the selection is made. The dictionary candidate acquisition process ends.

  When at least one of the rule ID stored in the first target memory Rule-ia and the rule ID stored in the second target memory Rule-ib is stored in the selected dictionary memory (YES in ST55), The dictionary selection unit 13 proceeds to information determination processing (FIG. 11: ST56, ST57, ST58) indicating the dictionary selection method stored in the dictionary selection method memory Dictionary.

  In the determination process, first, it is determined whether or not the information indicating the dictionary selection method stored in the dictionary selection method memory Dictionary is “Both” (FIG. 11: ST56). When the information indicating the dictionary selection method is “Both” (YES in ST56), the dictionary selection unit 13 is paired with the rule ID stored in the first target memory Rule-ia in the selected dictionary memory. Deleting the selected dictionary ID stored in the selected dictionary memory in a pair with the stored selection dictionary ID and the rule ID stored in the second target memory Rule-ib from the selected dictionary memory; The selected dictionary candidate memory is added (FIG. 11: ST59).

  When the information indicating the dictionary selection method is not “Both” (NO in ST56), the dictionary selection unit 13 determines whether or not the information indicating the dictionary selection method stored in the dictionary selection method memory Dictionary is “TrueOnly”. Judgment is made (FIG. 11: ST57). When the information indicating the dictionary selection method is “TrueOnly” (YES in ST57), the dictionary selection unit 13 is paired with the rule ID stored in the first target memory Rule-ia in the selected dictionary memory. If there is a stored selected dictionary ID, the selected dictionary ID is deleted from the selected dictionary memory and added to the selected dictionary candidate memory. If there is a selection dictionary ID stored in the selection dictionary memory in a pair with the rule ID stored in the second target memory Rule-ib, the selection dictionary ID is deleted from the selection dictionary memory. Then, it is added to the selected dictionary candidate memory (FIG. 11: ST60).

  If the information indicating the dictionary selection method is not “TrueOnly” (NO in ST57), the dictionary selection unit 13 determines whether the information indicating the dictionary selection method stored in the dictionary selection method memory Dictionary is “Rule2Only”. Is determined (FIG. 11: ST58). When the information indicating the dictionary selection method is “Rule2Only” (YES in ST58), the dictionary selection unit 13 is paired with the rule ID stored in the second target memory Rule-ib in the selected dictionary memory. The stored selected dictionary ID is deleted from the selected dictionary memory and added to the selected dictionary candidate memory (FIG. 11: ST61). This completes the selection dictionary candidate acquisition process.

  When the selected dictionary candidate acquisition process ends, the dictionary selection unit 13 adds the connection rule ID stored in the ID memory ID and the dictionary ID in the selected dictionary candidate memory to the selected dictionary memory (see FIG. 10: ST38). Thereafter, returning to ST33, the counter i is further counted up, and the processing after ST34 is executed again.

  Thus, when the processing of ST35 to ST38 is executed for all the linked dictionary rule data recorded in the dictionary linkage rule file 22, the counter i exceeds the threshold value N (YES in ST34), and the dictionary selection unit 13 Creates a selected dictionary list based on the dictionary ID in the selected dictionary memory (FIG. 10: ST39). This completes the rule connection process.

  When the rule connection process ends, the dictionary selection unit 13 transmits the metadata generated by the metadata generation unit 12 and the selected dictionary list created by the rule connection process to the dictionary update unit 14. Based on the dictionary update rule recorded in the dictionary update rule file 23, the dictionary update unit 14 executes a process for updating the dictionary file identified by the dictionary ID recorded in the selected dictionary list (FIG. 8: ST8). .

  First, the number N of dictionary update rule data recorded in the dictionary update rule file 23 is stored in the threshold memory (FIG. 12: ST71). Further, the count value i of the counter memory is reset to “0” (FIG. 12: ST72).

  When the processes of ST71 to ST72 are completed, the dictionary updating unit 14 increments the count value i of the counter memory by “1” (FIG. 12: ST73). Then, each time counting up, it is compared with the value N of the threshold memory (FIG. 12: ST74), and the processing of ST75 to ST89 is repeatedly executed until the count value i exceeds the threshold value N.

  That is, the dictionary update unit 14 reads i-th (i is a count value) dictionary update rule data from the dictionary update rule file 23 and stores it in the update rule memory UPRule-i (FIG. 12: ST75). Then, it is determined whether or not the dictionary ID of the dictionary update rule data stored in the update rule memory UPRule-i is included in the selected dictionary list (FIG. 12: ST76). If not included (NO in ST76), the process returns to ST73, further increments the counter i, and executes the processes after ST74 again.

  If the dictionary ID of the dictionary update rule data stored in the update rule memory UPRule-i is included in the selected dictionary list (YES in ST76), the dictionary update unit 14 updates the dictionary stored in the update rule memory UPRule-i. The external linkage ID of the rule data is checked (FIG. 12: ST77).

  When the external linkage ID is “0” (NO in ST77), the dictionary update rule data need not be linked with the external system 6. In this case, the dictionary update unit 14 recognizes the attribute of the dictionary update rule data stored in the update rule memory UPRule-i. Then, the value of this attribute is extracted from the metadata received from the dictionary selection unit 13 and stored in the first comparison memory mValue (FIG. 12: ST78). Further, the comparison value of the dictionary update rule data stored in the update rule memory UPRule-i is stored in the second comparison memory rValue (FIG. 12: ST79).

  On the other hand, when the external linkage ID of the dictionary update rule data stored in the update rule memory UPRule-i is “1” or more (YES in ST77), the dictionary update rule data is linked to the external system 6. In this case, the dictionary update unit 14 transmits the metadata received from the dictionary selection unit 13 and the external linkage ID of the dictionary update rule data to the related information search unit 15 (FIG. 12: ST80). Then, it waits for a return value from the related information search unit 15 (ST81 in FIG. 12). The operations of the related information search unit 15 and the external system linkage unit 16 at this time are the same as those in the dictionary selection process.

  When the dictionary update unit 14 receives a return value from the external system 6 via the related information search unit 15, the dictionary update unit 14 stores the return value in the return value memory sValue (FIG. 12: ST82). Next, the dictionary update unit 14 determines whether the target data code of the dictionary update rule data stored in the update rule memory UPRule-i is “1” or “2” (FIG. 12: ST83).

  When the target data code is “1” (YES in ST83), the dictionary updating unit 14 stores the value of the return value memory sValue in the first comparison memory mValue (FIG. 12: ST84). Further, the comparison value of the dictionary update rule data stored in the update rule memory UPRule-i is stored in the second comparison memory rValue (FIG. 12: ST85).

  When the target data code is “2” (NO in ST83), the dictionary update unit 14 acquires an attribute from the dictionary update rule data stored in the update rule memory UPRule-i. Then, a value corresponding to the attribute is extracted from the metadata received from the dictionary selection unit 13 and stored in the first comparison memory mValue (FIG. 12: ST86). Further, the value of the return value memory sValue is stored in the second comparison memory rValue (FIG. 12: ST87).

  When the values are stored in the first comparison memory mValue and the second comparison memory rValue, respectively, the dictionary updating unit 14 compares the value of the first comparison memory mValue and the value of the second comparison memory rValue. Then, it is determined whether or not the conditions of the dictionary update rule data stored in the update rule memory UPRule-i are satisfied (FIG. 12: ST88). If the condition is not satisfied (NO in ST88), the process returns to ST73, the counter i is further counted up, and the processes after ST74 are executed again.

  When the condition is satisfied (YES in ST88), the dictionary update unit 14 identifies the dictionary identified by the dictionary ID of the dictionary update rule data based on the update contents of the dictionary update rule data stored in the update rule memory UPRule-i. The file data is updated (FIG. 12: ST89). Thereafter, the process returns to ST73, the counter i is further counted up, and the processes after ST74 are executed again.

  Thus, when the processes of ST75 to ST89 are executed for all dictionary update rule data recorded in the dictionary update rule file 23, the counter i exceeds the threshold value N (YES in ST74), and the dictionary update process is performed. End (dictionary updating means).

  In the server 1 having the dictionary update function 7 having such a configuration, when an e-mail is received from the client 2, the input unit 11 extracts header information and a mail text as accompanying data from the e-mail. In addition, when an attached file other than an image file exists in the e-mail, text data included in the attached file is acquired. The accompanying data and text data are sent to the metadata generation unit 12.

  In the metadata generation unit 12, metadata is generated from the header information of the accompanying data and the mail text. If text data is included, metadata is also generated from this text data. The generated metadata is transmitted to the dictionary selection unit 13.

  The dictionary selection unit 13 determines whether or not a dictionary selection rule condition registered in the dictionary selection rule file 21 is satisfied based on the metadata. At the time of this determination, the dictionary selection rule set to cooperate with the external system 6 is determined in cooperation with the corresponding external system 6. When the condition is satisfied, the rule ID of the rule and the selected dictionary ID set in the rule are paired and stored in the selected dictionary memory.

  Next, the dictionary selection unit 13 determines whether or not the connection condition of the dictionary connection rule registered in the dictionary connection rule file 22 is satisfied for the rule ID stored in the selected dictionary memory in a pair with the selected dictionary ID. Is determined. For the two rule IDs that satisfy the connection condition, the selection dictionary ID paired with the rule ID according to the dictionary selection method of the connection rule remains in the selection dictionary memory.

  Thus, the dictionary selection unit 13 creates a selection dictionary list from the selection dictionary ID stored in the selection dictionary memory. The created selected dictionary list is sent to the dictionary update unit 14 together with the metadata generated by the metadata generation unit 12.

  In the dictionary update unit 14, a dictionary update rule for the selected dictionary ID registered in the selected dictionary list is selected from each dictionary update rule registered in the dictionary update rule file 23. Then, based on the metadata, it is determined whether or not the condition of the selected dictionary update rule is satisfied. In this determination, a dictionary update rule that is set to be linked with the external system 6 is determined in cooperation with the corresponding external system 6. When the condition is satisfied, the dictionary file identified by the dictionary ID of the rule is updated.

  As described above, according to the server 1 of the present embodiment, the recognition dictionary file can be automatically updated without imposing a burden on the user based on the data related to the dictionary update input to the input unit 11. . Therefore, labor and time required for maintenance of the recognition dictionary file can be saved, and the cost can be reduced.

(Second Embodiment)
Next, a second embodiment according to the character recognition device of the present invention will be described with reference to FIGS. In the second embodiment, parts that are the same as those in the first embodiment are given the same reference numerals as much as possible, and detailed descriptions thereof are omitted.

  FIG. 13 is a system configuration diagram of the second embodiment. This system also includes a server 100, a client 2, and a multifunction machine 3, and these are connected by a network 4.

  The server 1 functions as a dictionary update device and a character recognition device. That is, the server 1 uses the various recognition dictionary files registered in the dictionary database 5 in addition to the dictionary update function 7, and performs character recognition for performing character recognition of the image file received from the client 2 or the multifunction device 3. Function 101 is provided. The dictionary update function 7 and the character recognition function 101 are realized by a character recognition program installed in the server 100.

  The character recognition program may be stored in advance in a storage device inside the server 1 or may be downloaded to the server 1 from a network. Further, what is stored in the recording medium may be installed in the server 1. The recording medium may be in any form as long as it can store a program such as a CD-ROM and can be read by the server 1.

  FIG. 14 is a block diagram showing components necessary for realizing the dictionary update function 7 and the character recognition function 101 in the server 1. In the present embodiment, an image acquisition unit 111, a character recognition unit 112, and an output unit 113 are added as compared with the first embodiment shown in FIG. Other components 11 to 17 are the same as those in the first embodiment.

  FIG. 15 is a flowchart showing a basic processing procedure regarding the dictionary update function 7 and the character recognition function 101 of the server 1. Hereinafter, the function of each component of the server 1 will be described with reference to this flowchart as appropriate. Note that description of parts common to the first embodiment will be simplified.

  The input unit 11 waits for data related to dictionary updating to be input (FIG. 15: ST101). When data related to dictionary update is input via the network 4, the input unit 11 analyzes the input data (FIG. 15: ST102). Then, image data is extracted from the input data (FIG. 15: ST103). Further, the accompanying data is extracted from the input data (FIG. 15: ST104).

  The input unit 11 transmits image data and accompanying data extracted from the input data to the metadata generation unit 12. The metadata generation unit 12 generates metadata of the input data based on the accompanying data received from the input unit 11 (FIG. 15: ST105).

  Further, the input unit 11 transmits image data extracted from the input data to the image acquisition unit 111. When acquiring the image data from the input unit 11, the image acquisition unit 111 stores the number N of the image data in the threshold memory (FIG. 15: ST106). Also, the image data list memory is cleared (FIG. 15: ST107). Further, the count value i of the counter memory is reset to “0” (FIG. 15: ST108). The threshold memory, the image data list memory, and the counter memory are all formed in the internal RAM of the server 1.

  When the processes of ST101 to ST108 are completed, the image acquisition unit 111 increments the count value i of the counter memory by “1” (FIG. 15: ST109). Each time counting up, the value is compared with the value N in the threshold memory (FIG. 15: ST110), and the processes of ST111 to ST112 are repeated until the count value i exceeds the threshold value N.

  That is, the image acquisition unit 111 performs preprocessing such as noise removal and format conversion on the i-th image data acquired from the input unit 11 (FIG. 15: ST111). Then, the image data that has been subjected to the preprocessing is added to the image data list memory (FIG. 15: ST112).

  When the preprocessing is completed for all the image data acquired from the input unit 11, the count value i exceeds the threshold value N. If count value i exceeds threshold value N (YES in ST110), image acquisition unit 111 provides list data of images stored in the image data list memory to character recognition unit 112 (image acquisition means).

  On the other hand, the dictionary selection unit 13 selects one or more update target dictionary files based on the dictionary selection rules recorded in the dictionary selection rule file 21 (FIG. 15: ST113). When the dictionary selection process ends, the dictionary selection unit 13 next executes a rule connection process (FIG. 15: ST114). When the rule connection process ends, the dictionary selection unit 13 transmits the metadata generated by the metadata generation unit 12 and the selected dictionary list created by the rule connection process to the dictionary update unit 14 and the character recognition unit 112. To do.

  Based on the dictionary update rule recorded in the dictionary update rule file 23, the dictionary update unit 14 executes a process of updating the dictionary file identified by the dictionary ID recorded in the selected dictionary list (FIG. 15: ST115). .

  The character recognition unit 112 captures image data from the image data list. Then, the character recognition of the image data is performed using the dictionary file identified by the dictionary ID recorded in the selected dictionary list (FIG. 15: ST116). The output unit 113 attaches the character recognition result in the character recognition unit 112 to the input data and transmits it to the client 2 (FIG. 15: ST117). For example, the character recognition result is transmitted as text data to the client 2 together with the image data of the input data. Alternatively, the text data of the character recognition result is set in the attached information of the image data and transmitted to the client (character recognition means).

  As described above, according to the server 100 of the present embodiment, the recognition dictionary file can be automatically updated based on the data related to the dictionary update input to the input unit 11 without imposing a burden on the user. In addition, since the recognition dictionary file after the update can be used to perform character recognition of the image data attached to the input data, highly accurate character recognition can be realized for handwritten characters and the like.

The present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage.
For example, in the above embodiment, e-mail data and document reading data are exemplified as data relating to dictionary update. However, the present invention is not limited to this, and image data that is a character recognition target and data associated with the image data are not limited thereto. Can be used as data relating to dictionary update. That is, any device that can transmit data composed of the image data and accompanying data to the server 1 or the server 100 can be incorporated into the system of the present embodiment.

  In addition, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, the constituent elements over different embodiments may be combined.

  DESCRIPTION OF SYMBOLS 1 ... Server (dictionary update apparatus), 2 ... Client, 3 ... Multifunction device, 4 ... Network, 5 ... Dictionary database, 6 ... External system, 7 ... Dictionary update function, 11 ... Input part, 12 ... Metadata production | generation part, DESCRIPTION OF SYMBOLS 13 ... Dictionary selection part, 14 ... Dictionary update part, 15 ... Related information search part, 16 ... External system cooperation part, 20 ... Rule database, 21 ... Dictionary selection rule file, 22 ... Dictionary connection rule file, 23 ... Dictionary update rule File: 30 ... Cooperation information database, 100 ... Server (dictionary update device, character recognition device), 101 ... Character recognition function, 111 ... Image acquisition unit, 112 ... Character recognition unit, 113 ... Output unit.

JP-A-8-7032

Claims (6)

In a dictionary update device that updates a dictionary file used for character recognition processing,
Input means for extracting accompanying data accompanying the image data from the input image data;
Metadata generation means for generating metadata in which attributes and values of the accompanying data are set from the accompanying data;
An external system linkage means for giving an instruction defined in the dictionary selection rule to an external system defined in the dictionary selection rule and obtaining the result;
The dictionary selection rule is determined based on the information obtained from the external system by the external system cooperation unit and the metadata generated by the metadata generation unit, and the dictionary file defined in the rule established is selected. Dictionary selection means to
A dictionary update unit that determines a dictionary update rule set for the dictionary file selected by the dictionary selection unit, and updates the dictionary file set in the rule according to the update content defined in the established dictionary update rule; ,
A dictionary updating apparatus comprising:   The dictionary selection means determines a dictionary selection rule for each of a plurality of dictionary selection rules defined in the dictionary connection rule, and when the determination result satisfies the dictionary connection rule, the dictionary selection rule is defined in the dictionary connection rule. 2. The dictionary updating apparatus according to claim 1, wherein a dictionary file is selected. A computer that can access the dictionary file used for character recognition processing.
Input means for extracting accompanying data accompanying the image data from the input image data;
Metadata generation means for generating metadata in which attributes and values of the accompanying data are set from the accompanying data,
An external system linkage means for giving an instruction defined in the dictionary selection rule to the external system defined in the dictionary selection rule and obtaining the result;
The dictionary selection rule is determined based on the information obtained from the external system by the external system cooperation unit and the metadata generated by the metadata generation unit, and the dictionary file defined in the rule that is established is selected. The dictionary selection means and the dictionary update rule set for the dictionary file selected by the dictionary selection means are determined, and the dictionary file set in the rule is updated according to the update contents defined in the established dictionary update rule Dictionary updating means,
Dictionary update program to function as.   The dictionary selection means determines a dictionary selection rule for each of a plurality of dictionary selection rules defined in the dictionary connection rule, and when the determination result satisfies the dictionary connection rule, the dictionary selection rule is defined in the dictionary connection rule. The dictionary update program according to claim 1, which functions to select a dictionary file. In a character recognition device that performs character recognition processing using a dictionary file,
Input means for extracting accompanying data accompanying the image data from the input image data;
Metadata generation means for generating metadata in which attributes and values of the accompanying data are set from the accompanying data;
An external system linkage means for giving an instruction defined in the dictionary selection rule to an external system defined in the dictionary selection rule and obtaining the result;
The dictionary selection rule is determined based on the information obtained from the external system by the external system cooperation unit and the metadata generated by the metadata generation unit, and the dictionary file defined in the rule that is established is selected. Dictionary selection means;
A dictionary update unit that determines a dictionary update rule set for the dictionary file selected by the dictionary selection unit, and updates the dictionary file set in the rule according to the update content defined in the established dictionary update rule; ,
Image acquisition means for acquiring image data input via the input means;
Character recognition means for recognizing characters using the dictionary file updated by the dictionary update means, the image data obtained by the image acquisition means,
A character recognition device comprising: A computer that can access the dictionary file used for character recognition processing.
Input means for extracting accompanying data accompanying the image data from the input image data;
Metadata generation means for generating metadata in which attributes and values of the accompanying data are set from the accompanying data,
An external system linkage means for giving an instruction defined in the dictionary selection rule to the external system defined in the dictionary selection rule and obtaining the result;
The dictionary selection rule is determined based on the information obtained from the external system by the external system cooperation unit and the metadata generated by the metadata generation unit, and the dictionary file defined in the rule that is established is selected. Dictionary selection means,
A dictionary update unit that determines a dictionary update rule set for the dictionary file selected by the dictionary selection unit, and updates the dictionary file set in the rule according to the update content defined in the established dictionary update rule;
Image acquisition means for acquiring image data input through the input means, and character recognition means for recognizing the image data obtained by the image acquisition means using the dictionary file updated by the dictionary update means ,
Character recognition program to function as

Images