JP2017016549A - Character recognition device, character recognition method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To eliminate the need for an operation to confirm specific information on an apparatus by a user and manual input of inspection items by a user, thereby improving work efficiency.SOLUTION: Provided is a character recognition device comprising: a storage unit for storing a recognition dictionary in which a plurality of groups including a plurality of items to recognize characters, an order of recognition within a group of the plurality of items included in the group, and information used in the character recognition of each item are registered; and a processor. The items registered in the recognition dictionary include a plurality of first items and a plurality of second items, a first item being the item that appears first in the order of recognition within a group, at least one second item being associated with the first item. The processor recognizes the character of a first item from an inputted image on the basis of information used for recognizing the characters of the first items of the plurality of groups registered in the recognition dictionary, specifies a second item to recognize a character next from a second item associated with the first item of the recognized character, and recognizes the character of the second item from the inputted image on the basis of information used for recognizing the character of the specified second item, the information being read out from the recognition dictionary.SELECTED DRAWING: Figure 5

Description

  The present invention relates to a character recognition device that recognizes characters from an input image.

  The number of information communication devices and the amount of data handled by the information communication devices are increasing dramatically. In order to operate an information service using an information communication device or the like, it is necessary to operate the information communication device or the like stably. For this reason, the importance of inspection work of information communication devices and the like is increasing in places where the information communication devices and the like are installed.

In the conventional inspection work, an operator manually writes the inspection result on a paper inspection table. For this reason, there are the following problems in terms of operational efficiency of inspection work and accuracy of inspection work.
(1) High cost due to the use of paper: A separate punch input operation is required to convert the inspection results entered in the inspection table into data, which increases costs.
(2) Complexity of confirmation work of model number: In the inspection work, after confirming the model number of the device to be inspected, the worker confirms the inspection item corresponding to the device. However, in places where many devices with similar model numbers are installed, the characters of the model numbers and inspection items are small and the installation location is likely to be dark. It is complicated to confirm the number or code of the item. In such a situation, the visual confirmation of the model number by the worker takes time and is likely to be erroneous.
(3) Complexity of the inspection work: In the inspection work, the items to be inspected are filled in, and after checking one item, the worker looks at the paper and confirms the next item, The number or code or the like of the item is found and the number or code is entered in the inspection table, but this inspection work is complicated.

  With respect to the problems (1) to (3), there is a technique described in Japanese Patent Laid-Open No. 2015-1851 (Patent Document 1). In the gazette of Patent Document 1, “a first extraction unit that extracts identification information of equipment or equipment scheduled to carry out the same type of inspection, equipment equipment name information, inspection content information, support information, and related equipment information; A second extraction unit for extracting hierarchical information and location specifying information corresponding to the facility or equipment to be inspected, a candidate specifying unit for specifying a candidate for the facility or equipment to be inspected next, and an inspection. A display generation unit that generates inspection scenario display information for displaying identification information, inspection content information, support information, and candidates corresponding to the facility or equipment to be implemented on the display unit of the terminal used by the worker; It is an inspection support device that is provided. ”(See summary).

Japanese Patent Laying-Open No. 2015-1851

  In the technique described in Patent Document 1, since the number or code of the inspection item is manually input as data to the terminal by the worker, the problem (1) is solved. Further, since the inspection scenario is displayed on the operator's terminal, the problem (3) is solved.

  However, in the technique described in Patent Document 1, the worker must confirm the model number and the like, and the worker must manually enter the numbers or codes of the inspection items. For this reason, the problem (2) is not solved. In addition, it is necessary for the worker to search for the inspection item.

  The present invention recognizes the character of the device specific information and recognizes the character of the check item corresponding to the device specific information, so that the user can confirm the device specific information and the user can manually check the check item. An object of the present invention is to provide a character recognition device that eliminates the need for input and improves work efficiency.

  A typical example of the present invention is a character recognition device that recognizes characters included in an input image, and includes a plurality of groups including a plurality of items for recognizing characters, and a plurality of groups included in the group. A storage unit that stores a recognition dictionary in which recognition order of items within the group and information used for character recognition of each item are registered, and a processor, and the items registered in the recognition dictionary are: Including a plurality of first items and a plurality of second items, wherein the first item is the first item in the recognition order in the group, and at least one second item is associated with the first item, The processor recognizes the character of the first item from the input image based on information used for recognizing the character of the first item of the plurality of groups registered in the recognition dictionary, and is registered in the recognition dictionary. Recognition order The second item for recognizing the character next is specified from the second item associated with the first item of the recognized character, and the information used for recognizing the character of the specified second item is recognized. Characters of the second item are recognized from the input image based on information used for recognizing the character of the second item read out from the dictionary.

  The effects obtained by the representative ones of the inventions disclosed in the present application will be briefly described as follows. That is, it is possible to provide a character recognition device that improves the work efficiency by eliminating the need for the user to check the device specific information and manually input the inspection items of the user.

  Problems, configurations, and effects other than those described above will be clarified by the following description of embodiments.

FIG. 6 is an explanatory diagram of an example of inspection work according to the first embodiment. 1 is a hardware configuration diagram of a character recognition device according to Embodiment 1. FIG. It is explanatory drawing of the inspection scenario data of Example 1. FIG. It is explanatory drawing of the work log of Example 1. FIG. It is a flowchart of the character recognition process performed by the character recognition part of Example 1. FIG. It is a flowchart of the model number collation process and item recognition process of Example 1. It is a flowchart of the character recognition pre-process of Example 1, and a character string recognition process. 3 is an explanatory diagram of a date dictionary according to Embodiment 1. FIG. It is explanatory drawing of the place selection reception screen of Example 1. FIG. It is explanatory drawing of the model number list display screen of Example 1. FIG. It is explanatory drawing of the recognition result display screen of the model number of Example 1. FIG. It is explanatory drawing of the inspection item list display screen of Example 1. FIG. It is explanatory drawing of the recognition result display screen of S / N of Example 1. FIG. 10 is a flowchart of character recognition processing according to the second embodiment. It is a flowchart of the item batch recognition process of Example 2. 6 is a flowchart of an item batch confirmation process according to the first embodiment. It is explanatory drawing of the recognition result display screen of the model number of Example 2. FIG. It is explanatory drawing of the recognition result display screen of the item (S / N) of Example 2. FIG. It is explanatory drawing of the recognition result display screen of the item (MAC) of Example 2. FIG. It is explanatory drawing of the unrecognized item display screen of Example 2. FIG. 14 is a flowchart of character recognition processing according to the third embodiment. It is explanatory drawing of the recognition result display screen which displays the list of the inspection items of Example 3. FIG. It is explanatory drawing of the recognition result display screen of the model number of Example 3. FIG.

  Example 1 will be described with reference to FIGS.

  In the present embodiment, for example, a character recognition device used for an inspection operation of information equipment (for example, a server or the like) installed in a data center or various stores will be described. However, the character recognition device according to the present embodiment is not limited to the inspection work, and can be applied to a work for recognizing characters from images and moving images.

  First, an outline of the present embodiment will be described.

  The character recognition device of the present embodiment automatically recognizes item characters that need to be recognized from the input image, and stores the recognized characters as the inspection work result. As a result, the user's work efficiency can be improved.

  FIG. 1 is an explanatory diagram of an example of inspection work according to the first embodiment.

  In FIG. 1, as an example, the portable terminal 100 will be described as a character recognition device having a character recognition unit 251 (see FIG. 2).

  A user (for example, a worker who performs inspection work) photographs the label 141 attached to the device 140 with the camera of the mobile terminal 100 (for example, a tablet terminal or a smartphone). An image captured by the camera is displayed as a recognition result display screen 110 on the display of the mobile terminal 100. The recognition result display screen 110 includes a captured image display area 120 and a character recognition result display area 130. In the captured image display area 120, an image captured by a camera is displayed, and a character string in which characters are recognized among character strings included in the image is surrounded and displayed by a rectangle 121.

  The character recognition result display area 130 includes a recognized character string enlarged image display unit 131, a character recognition result display unit 132, an OK button 133, and a Retry button 134.

  The recognized character string enlarged image display unit 131 displays an enlarged image of the character string in which the character is recognized. The recognized character is displayed on the character recognition result display unit 132. The recognized character string enlarged image display unit 131 and the character recognition result display unit 132 allow the user to compare the recognized character string with the recognized character string.

  When the user determines that the recognized character is correct, the user operates the OK button 133. On the other hand, when the user determines that the recognized character is not correct, the user operates the Retry button 134 to recognize the character again or inputs the correct character manually.

  In the present embodiment, in order to improve the accuracy of character recognition, characters are recognized using the inspection scenario data 261 (see FIG. 2) and the work log 262 (see FIG. 2).

(1) Character recognition based on inspection scenario data 261 The inspection scenario data 261 is described as a recognition dictionary in the claims, and the inspection scenario data 261 includes a plurality of groups including a plurality of items for recognizing characters, The recognition order of items in the group and information used for character recognition of each item are registered. The items registered in the inspection scenario data 261 include a plurality of first items and a plurality of second items, and the first item is the first item in the recognition order in the group, and identifies the device. Device identification information (for example, model number). In addition, at least one second item is associated with the first item. The second item is an item to be inspected other than the device identification information.

  The portable terminal 100 recognizes characters of each item by switching a dictionary to be used for each item to be checked. Thereby, since the portable terminal 100 recognizes the character of each item using the dictionary suitable for the check item, the accuracy of character recognition can be improved.

(2) Character recognition based on work log 262 In the work log 262, the recognition result of each item and various parameters for correcting an image are registered. A recognition result and various parameters when each item of each device is inspected are accumulated in the work log 262. When an item for which character recognition has been successful once is recognized again, the work log 262 for which character recognition has been successful can be used, and the accuracy of character recognition for the item can be improved.

  By using the inspection scenario data 261 and the work log 262 of this embodiment for character recognition, the accuracy of character recognition can be improved and the inspection work can be made more efficient. The work log 262 is not necessarily used for character recognition.

  For example, information used for character recognition of each item of the inspection scenario data 261 includes an item character type, an item character number, an item notation format, and the like. The character type is a character type (for example, a number, an alphabet, a kana, a kanji, or a symbol) used for the item character. The number of characters is the number of characters of the item, and may be the number of digits if the character of the item is a number. The notation format is a character-specific character notation format. For example, the date format is “YY.MM.DD” or “YY / MM / DD”. Here, “YY”, “MM”, and “DD” mean numbers representing year, month, and day, respectively. For example, in the case of “MM”, the numbers are “1-12” and “01-09”.

  In this embodiment, the recognition dictionary is switched corresponding to the item for recognizing characters. For example, if the date “15.01.18” in the image is recognized without using the recognition dictionary, it may be erroneously recognized as “l (el) 5.O1 · old”. “1” in the image is misrecognized by the alphabet “El”, “0” in the image is misrecognized by the alphabet “O”, “.” In the image is misrecognized by the middle point, In “18”, “1” and “8” are recognized as one character and are erroneously recognized as “old”.

  For example, if a character type is a number and a notation format “YY.MM.DD” is registered as a date item recognition dictionary, such misrecognition can be prevented, and characters of the date item can be prevented. Can be accurately recognized.

  Further, since the inspection order of items is registered in the inspection scenario data 261, the recognition dictionary can be automatically switched in the inspection order of the inspection work, and the inspection scenario data 261 appropriate for the item for recognizing characters is used. Can recognize characters.

  As described above, the user simply captures the label 141 of the device 140 with the camera of the mobile terminal 100, and the characters of items that need to be input in the inspection work are input to the mobile terminal 100. Thereby, a user's inspection efficiency can be improved.

  Note that the mobile terminal 100 is not necessarily a character recognition device having the character recognition unit 251. For example, the mobile terminal 100 transmits an image captured by itself to a server (character recognition device) having the character recognition unit 251, and the server executes character recognition processing and transmits the character recognition result to the mobile terminal 100. . Then, the mobile terminal 100 may display a character recognition result or the like on the display of the mobile terminal 100.

  FIG. 2 is a hardware configuration diagram of the character recognition device 200 according to the first embodiment.

  As described above, the character recognition device 200 may be the mobile terminal 100 or a server to which the mobile terminal 100 can be connected via a network.

  The character recognition device 200 includes an operation unit 210, a display unit 220, an image input unit 230, a processor 250, a memory (main storage device) 260, an auxiliary storage device 270, and an interface (IF) 280.

  The operation unit 210 receives an input of an operation by a user of the character recognition device 200. The operation unit 210 is, for example, a touch panel, a keyboard, a mouse, or a button. The display unit 220 is a display or the like that displays an image input to the character recognition device 200, a character recognition result, and the like. The image input unit 230 is a digital camera, a scanner, or the like that captures an image using an image sensor such as a CCD (Charge Coupled Device).

  The processor 250 executes a program stored in the memory 260. The memory 260 includes a ROM that is a nonvolatile storage element and a RAM that is a volatile storage element. The ROM stores an immutable program (for example, BIOS). The RAM is a high-speed and volatile storage element such as a DRAM (Dynamic Random Access Memory), and temporarily stores a program executed by the processor 250 and data used when the program is executed.

  The auxiliary storage device 270 is a large-capacity non-volatile storage device such as a magnetic storage device (HDD) or a flash memory (SSD), and stores a program executed by the processor 250 and data used when the program is executed. To do. That is, the program is loaded from the auxiliary storage device 270 to the memory 260 and executed by the processor 250.

  The processor 250 includes a character recognition unit 251 that recognizes characters in the input image. The character recognition unit 251 includes a device identification information recognition unit 252 and an inspection item recognition unit 253. The device identification information recognition unit 252 recognizes the characters of the device identification information (first item) from the input image. The inspection item recognition unit 253 recognizes a plurality of second item characters associated with the recognized first item from the input image. That is, the inspection item recognition unit 253 recognizes characters of items to be inspected with the device specified by the device specifying information that is the first item. The memory 260 stores a program corresponding to the character recognition unit 251, and the character recognition unit 251 is mounted on the processor 250 when the processor 250 executes the program.

  In the memory 260, inspection scenario data 261 and work log 262 are stored. In the inspection scenario data 261, the order of items for recognizing characters and a recognition dictionary used for recognizing characters of each item are registered for each type of device. Details of the inspection scenario data 261 will be described with reference to FIG. In the work log 262, the character recognition result of each item and various parameters for adjusting the image are registered. Details of the work log 262 will be described with reference to FIG.

  The interface 280 is an interface that connects the character recognition device 200 to a network or other devices.

  The image may be acquired from the interface 280 instead of being captured by the image input unit 230. Specifically, the image is stored in the portable storage medium, and the character recognition device 200 is connected to the portable storage medium via the interface 280 and recognizes the image that recognizes the character stored in the portable storage medium. You may get it. Further, an image may be stored in a server or the like connected to the character recognition device 200 via a network, and the character recognition device 200 may acquire an image from the server or the like via an interface.

  FIG. 3 is an explanatory diagram of the inspection scenario data 261 according to the first embodiment.

  The inspection scenario data 261 includes a location 301, a device number 302, a model number 303, an item number 304, an attribute 305, a recognition mode 306, a character recognition dictionary 307, a word dictionary 308, and a notation dictionary 309.

  In the place 301, an identifier of the place where the device to be inspected is installed is registered. For example, “A” registered in the location 301 is an identifier of the data center A, and “B” is an identifier of the data center B. In the device number 302, a number for identifying a device to be inspected is registered. In the model number 303, the model number of the device to be inspected is registered. In the item number 304, a number for identifying an item to be inspected corresponding to the device to be inspected is registered. In the attribute 305, an attribute of an item to be inspected is registered.

  In the recognition mode 306, a character recognition mode of an item to be checked is registered. For example, the recognition mode includes a collation mode, a reading mode, a selection mode, and a direct input mode. In the collation mode, when the recognized character matches the value registered in the word dictionary 308, the recognized character is set as a character recognition result, and when the recognized character does not match the value registered in the word dictionary 308, This mode discards recognized characters. The reading mode is a mode for reading recognized characters. The selection mode is a mode that accepts input of at least one option from a plurality of options displayed on the display unit 220. The direct input mode is a mode in which a user directly inputs characters and numbers. The collation mode and the reading mode are modes in which the character recognition device 200 actually recognizes characters from the image, and the selection mode and the direct input mode are modes in which character recognition by the character recognition device 200 is unnecessary.

  In the character recognition dictionary 307, the character type of the item is registered. For example, an item in which “English, number” is registered in the character recognition dictionary 307 means that the item is composed of alphabets and numbers. When recognizing the character of the item, only the alphabet character recognition dictionary and the numeric character recognition dictionary are used.

  In the word dictionary 308, words (values) with which recognized characters must match are registered. In the notation dictionary 309, the character notation format of the item is registered. For example, when “NNN-NNNN-JPNN” is registered in the notation dictionary 309, the characters of the recognized items are, in order from the left, three-digit numbers, “−”, four-digit numbers, “−” It means that the format must be “JP” and a two-digit number.

  The character recognition dictionary 307, the word dictionary 308, and the notation dictionary 309 are information used for character recognition.

  As described above, in the inspection scenario data 261, the items to be inspected, the inspection order of the items, and information used for recognition of the items are registered for each device to be inspected. In the record of the data center A shown in FIG. 3, there are a group of items to be inspected by a device having a model number “967-7363-JP03” and a group of items to be inspected by a device having a model number “AZ513-88E”. In the device having the model number “967-7363-JP03”, the “model number” is recognized first, then “S / N (serial number)” is recognized, and then “MAC address” is recognized. In the device having the model number “AZ513-88E”, the “model number” is recognized first, and then the “number of processes” is recognized.

  Since items to be inspected for each device are registered in the inspection scenario data 261, it is necessary to first recognize the model number and specify the device to be inspected. For this reason, in any group of the inspection scenario data 261, the model number is registered as an item to be inspected first.

  FIG. 4 is an explanatory diagram of the work log 262 according to the first embodiment.

  The work log 262 includes a work date 401, a place 402, a device number 403, a model number 404, an item number 405, an attribute 406, a recognition mode 407, a result value 408, a parameter A 409, and a parameter B 410.

  In the work day 401, the date on which the inspection work was performed is registered. Since the location 402 to the recognition mode 407 are the same as the location 301 to the recognition mode 306 of the inspection scenario data 261, description thereof is omitted. In the result value 408, the character recognition result of the item is registered. Parameters for correcting an image are registered in the parameters A409 and B410. The parameters for correcting the image will be described in detail with reference to FIG. In the parameters A409 and B410, the time required for character recognition, the total number of character recognition executions, the number of successful character recognitions, and the like may be registered. For example, items that require more than a predetermined value for character recognition and items whose ratio of the number of successes to the total number of executions of character recognition is less than or equal to a predetermined value are checked so as to prompt user input without character recognition. The scenario data 261 may be changed. In addition, an image used for character recognition may be registered in the parameter A409 and the parameter B410. The use of the image used for character recognition makes it possible to investigate whether or not the character recognition result is successful.

  FIG. 5 is a flowchart of character recognition processing executed by the character recognition unit 251 according to the first embodiment.

  The character recognition process includes three steps (1) to (3).

(1) Inspection location selection acceptance step (501)
The character recognition unit 251 accepts selection of a place where the inspection target device is installed.

(2) Device identification step (520)
After accepting the selection of the place, the device identification information recognition unit 252 of the character recognition unit 251 recognizes the character that identifies the device to be inspected and identifies the device to be inspected.

(3) Inspection item recognition step (530)
After the device to be inspected is specified, the inspection item recognition unit 253 of the character recognition unit 251 recognizes the character of the inspection item of the device specified in the device specifying step based on the inspection scenario data 261.

  Details of the steps (1) to (3) will be described below.

(1) Inspection location selection acceptance step (501)
In step 501, the character recognizing unit 251 displays a place selection reception screen 900 shown in FIG. 9A on the display unit 220 and receives an input of inspection place selection from the user. On the place selection reception screen 900, the name of the place identified by the place identifier registered in the place 301 of the inspection scenario data 261 is displayed. When the character recognition unit 251 receives an input of location selection, the character recognition unit 251 may refer to the inspection scenario data 261 and display on the display unit 220 a list of model numbers of devices installed at the location where the input is received. Specifically, the character recognition unit 251 acquires the model number registered in the model number 303 of the record in which the identifier of the place where the input is received in the place 301 is registered among the records registered in the inspection scenario data 261. The acquired model number is displayed on the display unit 220.

(2) Device identification step (520)
Next, the character recognizing unit 251 calls the device specifying information recognizing unit 252 to execute the device specifying step (520). The device identification information recognizing unit 252 reads a record in which the identifier of the place whose input is received in the process of step 501 is registered in the place 301 among the records registered in the inspection scenario data 261 (502).

  Next, the device identification information recognizing unit 252 reads a record in which the identifier of the place that has received the input in the process of step 501 is registered in the place 402 among the records registered in the work log 262 (503).

  Next, the device identification information recognizing unit 252 executes the model number matching process using the inspection scenario data 261 read in the process of step 502 and the work log 262 read in the process of step 503 (504). In the model number verification process, the device identification information recognition unit 252 recognizes the model number of the device to be inspected as device identification information from the input image. Details of the model number matching process will be described with reference to FIGS.

(3) Inspection item recognition step (530)
Next, the character recognition unit 251 calls the inspection item recognition unit 253 in order to execute the inspection item recognition step (530). Of the records registered in the inspection scenario data 261, the inspection item recognizing unit 253 registers the identifier of the place where the input was received in the process of step 501 in the place 301, and the model number recognized in the process of step 504 in the model number 303. Is read and the record of the item to be recognized next is read (505). For example, when the inspection item recognition step is executed immediately after the model number is recognized in the process of step 504, the item number recognized next to the model number is registered as the record to be recognized as the next item. Record. In the record in which “967-7363-JP03” is registered in the model number 303 of the inspection scenario data 261 illustrated in FIG. 3, the record in which “2” is registered in the item number 304 is read.

  Next, of the records registered in the work log 262, the inspection item recognizing unit 253 registers the identifier of the place that has received the input in the process of step 501 in the place 402, and recognizes the identifier of the model 404 in the process of step 504. The record of the item to be recognized next is read (506).

  Next, the inspection item recognition unit 253 executes an item recognition process using the inspection scenario data 261 read in the process of step 505 and the work log 262 read in the process of step 506 (507). In the item recognition process, the inspection item recognition unit 253 recognizes characters of items other than the model number from the input image. Details of the item recognition processing will be described with reference to FIGS.

  Next, the character recognition unit 251 determines whether or not the item recognition process has been executed for all items other than the model number corresponding to the model number recognized in the process of step 504 (508).

  In the process of step 508, when it is determined that the item recognition process has not been executed for all items other than the model number corresponding to the model number recognized in the process of step 504 (508: NO), the inspection item recognition unit 253 refers to the inspection scenario data 261 and determines an item to be recognized next (509). And the inspection item recognition part 253 returns to the process of step 505, reads the record of the item recognized next, and performs an item recognition process.

  For example, in the record in which “967-7363-JP03” is registered in the model number 303 of the inspection scenario data 261 illustrated in FIG. 3, when the item of the record in which “2” is registered in the item number 304 is recognized, The item to be recognized is a record in which “3” is registered in the item number 304.

  On the other hand, if it is determined in step 508 that the item recognition process has been executed for all items other than the model number corresponding to the model number recognized in step 504 (508: YES), the character recognition unit 251. Determines whether the processing in steps 520 and 530 has been executed for all the devices installed in the place where the input was received in the processing in step 501 (510).

  In the process of step 510, if it is determined that the processes of steps 520 and 530 have not been executed for all the devices installed at the place where the input was received in the process of step 501 (510: NO), character recognition The unit 251 determines the next device (511), and returns to the process of step 502.

  On the other hand, if it is determined in step 510 that the processes in steps 520 and 530 have been executed for all devices installed in the place where the input was received in step 501 (510: YES), character recognition is performed. The process ends.

  FIG. 6 is a flowchart of the model number matching process and the item recognition process according to the first embodiment.

  The model number matching process is executed by the device identification information recognizing unit 252 and the item recognition process is executed by the inspection item recognizing unit 253. In FIG. 6, the model number matching process and the item recognition process are described together. The subject will be described as the character recognition unit 251.

  First, the character recognition unit 251 refers to the inspection scenario data 261 and determines whether or not the item to be inspected requires character recognition (601). Items that do not require character recognition include, for example, items that only require a check mark indicating that they have been inspected, items that select at least one option from a plurality of options, and characters or numbers directly. There are items to enter. In the record recognition mode 306 of the item that does not require character recognition of the inspection scenario data 261, information (for example, a selection mode) indicating that fact is registered.

  If it is determined in step 601 that the item to be checked does not require character recognition (601: NO), the character recognition unit 251 accepts data input by the user (602), and proceeds to step 605. .

  On the other hand, if it is determined in step 601 that the item to be checked requires character recognition (601: YES), the character recognition unit 251 performs pre-character recognition processing (603). In the character recognition pre-processing, the character recognition unit 251 adjusts the input image and extracts a character string from the input image. Details of the character recognition pre-processing will be described with reference to FIG.

  Next, the character recognition unit 251 executes a character string recognition process (604). In the character string recognition process, the character recognition unit 251 recognizes the character string of the item to be inspected from the character string extracted in the process of step 603. Details of the character string recognition processing will be described with reference to FIG.

  Next, the character recognizing unit 251 presents the recognition result of the processing in step 604 to the user (605). The user determines whether or not the recognition result of the process in step 604 is correct based on the content presented in the process of step 605. When the user determines that the recognition result of the process at step 604 is incorrect, the user causes the character recognition unit 251 to execute the processes at steps 603 and 604 again or inputs correction contents to the character recognition apparatus 200.

  Next, the character recognition unit 251 registers the recognition result of the process in step 604 and the parameters used in the process of step 603 in the work log 262 (606), and ends the model number matching process or the item recognition process.

  FIG. 7 is a flowchart of pre-character recognition processing and character string recognition processing according to the first embodiment.

  In the character recognition pre-processing and the character string recognition processing, the character recognition unit 251 recognizes the character of the item to be checked while referring to the work log 262. In the work log 262, various parameters and recognition results when the character recognition is successful are registered. However, when character recognition fails, a parameter or a recognition result may be directly input to the work log 262 by a user or the like.

  First, an image is input from the image input unit 230 to the character recognition unit 251 (701). The character recognizing unit 251 refers to the focus determination parameter included in the parameters of the work log 262 and determines whether or not the image input in the process of step 701 is in focus. For example, the focus determination parameter is a density value distribution of an image for which character recognition has been successfully performed, and the character recognition unit 251 compares the density value distribution with the density value distribution of the input image. It is determined whether the image is in focus. If the character recognition unit 251 determines that the input image is not in focus, the character recognition unit 251 may present a screen prompting the user to input a new image.

  Next, the character recognizing unit 251 refers to the presence / absence of black / white reversal and the hue / luminance correction parameters included in the parameters of the work log 262, and corrects the image input in the processing of step 701 (702). In the inspection work, there is a high possibility that an image is captured by a camera. For example, the image quality of an image captured by a scanner that shoots light with the image sensor and paper in close contact with each other is stable, but the image quality of the image captured by the camera is easily affected by the external environment and unstable. It is. Examples of image correction include correction of contrast of an image captured in a dark place, and black and white inversion of an image including black and white reversal characters. If the parameter of the item to be inspected in the work log 262 is black and white reversal, the character recognition unit 251 reverses the black and white (negative / positive) of the image input in the process of step 701. Further, the character recognition unit 251 corrects the contrast of the image input in the process of step 701 based on the hue / brightness correction parameter included in the parameter of the item to be checked in the work log 262.

  Next, the character recognition unit 251 refers to the binarization method and the binarization parameter included in the parameters of the work log 262, and binarizes the image input in the process of step 701 (703). Generally, a color image or a gray image is converted into a black and white binary image, and characters in the converted binary image are recognized. For conversion to an appropriate binary image, an appropriate binarization method and binarization parameters must be used in accordance with the situation of the image input in the process of step 701. For example, appropriate binarization schemes and binarization parameters must be used for images with small differences between character color and background color, images with low contrast, and images with light text and dark background, respectively. . The character recognizing unit 251 stores the binarization method and binarization parameter of the item for which character recognition was successful, and when recognizing the character of the item, the binarization method and binarization parameter. Is used to convert a color image or a gray image into a binary image.

  Next, the character recognizing unit 251 refers to the shooting direction and inclination included in the parameters of the work log 262 and corrects the projection distortion, barrel distortion, inclination, etc. of the image input in the processing of Step 701 ( 704). Depending on the installation location of the device, the label to be read cannot be photographed from the front by the camera, and the label must be photographed from an oblique direction, or the character may be photographed in an inclined state. In the process of step 704, distortion or inclination of the image thus captured is corrected. In the work log 262, the positional relationship between the camera and the reading target (for example, the shooting direction) and the inclination of the character string at the time of shooting are registered. The character recognizing unit 251 corrects image distortion and inclination using the work log 262, and thus can correct image distortion and inclination accurately and at high speed.

  Next, the character recognizing unit 251 refers to the layout between items, the distance between character strings, and the distance between characters included in the parameters of the work log 262, and determines the character string of the item to be checked from the image in the process of step 701. Extract (705). A connected component fusion method is known as a character string extraction method. The connected component fusion method is a method of extracting a region where black pixels are continuous in a binary image as a connected component and extracting a character string from the arrangement relationship of the connected components. Specifically, a character string is extracted in consideration of the vertical or horizontal alignment of connected components adjacent to each other.

  There is a possibility that the image input in the processing of step 701 includes a character string other than the item to be checked. For this reason, the character recognizing unit 251 selects a character string of an item to be checked from the image input in the process of step 701. There are three methods for selecting the character string.

  The first method is to select only the character string of the item to be checked. In this embodiment, the user takes an image with the camera for each item to be inspected. In this case, the user captures an image so that the character string of the item to be checked is located near the center of the image. The character recognition unit 251 extracts a character string from the region near the center of the image input in the process of step 701 using a connected component fusion method. Thereby, the character recognition unit 251 can extract only the character string of the item to be checked.

  The second method is a method of extracting all character strings of the image input in the process of step 701 and selecting a character string of items to be checked in the character string recognition process which is the process of step 604. Specifically, in the character string recognition process, a character string that matches the word dictionary 308 and the notation dictionary 309 of the items to be checked among all the extracted character strings is selected as an item to be checked.

  The third method is a method in which the image input in the process of step 701 is displayed on the display unit 220, and designation of the character string range of the item to be checked is received from the user. Examples of the user designation method include a method in which only the center of the character string is designated, a method in which both ends of the character string are designated, and a method in which the four corners of the character string are designated.

  In the process of step 705, the character recognition unit 251 refers to the distance between character strings and the character string distance included in the work log 262 and extracts a character string from the image. In the first method, if the distance between the character string of the item to be inspected and the character string other than the item to be inspected is smaller than a predetermined value compared to the distance between the character strings, the two character strings are mistakenly regarded as one character string. There is a possibility to extract. If the distance between characters of the item to be checked is larger than the distance between characters in the work log 262, only a part of the character string of the item to be checked may be extracted.

  Next, the character recognizing unit 251 refers to the character recognition dictionary 711 and recognizes individual characters in the character string extracted in the process of step 705 (706). As a character recognition method, there is a modified quadrature discriminant function (MQDF). In this method, when a character recognition dictionary 711 in which a large number of characters are registered is generated, an identification function for calculating a distance scale is calculated for each character category from Bayes' theorem. When the input character is recognized, the input character is recognized by classifying the input character into the character category based on the distance scale output by the identification function. Note that a character recognition score indicating the probability of character recognition may be calculated based on the distance scale.

  In the processing of step 706, the character recognition unit 251 acquires the character type registered in the character recognition dictionary 307 of the record of the item to be inspected in the inspection scenario data, and stores the character recognition dictionary 711 corresponding to the acquired character type. Refer to and recognize characters. Thereby, the accuracy of character recognition can be improved.

  Next, the character recognition unit 251 performs a word matching process and a notation matching process on the character recognition result in the process of step 706 to calculate the final character recognition result of the item (707). The word collating process is a process of collating whether or not the character recognition result in the process of step 706 matches the word registered in the word dictionary 308 of the item record to be inspected in the inspection scenario data 261. Further, the notation collation process collates whether or not the notation format of the character recognition result in the processing of step 706 matches the notation format registered in the notation dictionary 309 of the record of the item to be inspected in the inspection scenario data 261. It is processing.

  A specific example of word matching processing will be described. For example, when two words “top” and “base” are registered in the word dictionary 308 of the item to be checked, and the character recognition result in the process of step 706 is “bass”, the character recognition unit 251 performs character recognition. It is determined that the result and the word registered in the word dictionary 308 do not match.

  The character recognition unit 251 corrects the character recognition result “bass” to “base” because the character recognition result “bass” and the “base” registered in the word dictionary 308 are one character different from each other. It may be calculated as typical character recognition. In the word matching process, the similarity (word matching score) between the character recognition result in the process of step 706 and the word registered in the word dictionary 308 is calculated, and the calculated similarity is similar to the predetermined value. , The word registered in the word dictionary 308 may be calculated as the final character recognition result.

  A specific example of the notation matching process will be described. When the character of the item to be inspected is written according to a certain rule, the notation checking process checks whether or not the character recognition result in step 706 matches the rule. For example, when an item to be checked is a date, a notation format such as “YY.MM.DD” or “YY / MM / DD” is registered in the notation dictionary 309 of the item, It is checked whether or not the notation format of the character recognition result in the process of step 706 matches the notation format such as “YY.MM.DD” or “YY / MM / DD”. In the notation matching process, it may be determined whether or not the character recognition result in the process of step 706 is within the numerical value or alphabetic range registered in the work log 262 or the notation dictionary 309.

  In the notation matching process, the similarity (notation matching score) between the notation format of the character recognition result in the process of step 706 and the notation format registered in the notation dictionary 309 may be calculated.

  Among the modes registered in the recognition mode 306 of the inspection scenario data 261, the modes that require character recognition include “reading mode” and “collation mode”. “Reading mode” is a mode for recognizing characters in an image. For example, the recognition mode of an item whose character or numerical value is variable is “reading mode”. The “collation mode” is a mode for collating whether or not the word registered in the word dictionary 308 matches the character recognition result. The recognition mode of an item whose character or numerical value is determined is “collation mode”. For example, since the “model number”, “S / N”, and “MAC address” have fixed characters or numerical values, the recognition mode of these items is “collation mode”, and the “processing number” has a variable numerical value. Therefore, the recognition mode is “reading mode”.

  In the “collation mode” item, word collation processing and notation collation processing are executed. However, since the word dictionary 308 is not registered in the “reading mode” item, only the notation collation processing is performed in the “reading mode” item. Executed. In this embodiment, the recognition mode of the item of the model number is “collation mode”, and the word collation process and the notation collation process are executed. However, the present invention is not limited to this. For example, the recognition mode of the item of the model number may be “reading mode”, and only the notation matching process may be executed. However, in order to improve the recognition accuracy, it is assumed that the characters of the item of the model number are recognized using at least the notation dictionary 309. For items other than the model number, an appropriate recognition mode is set according to the specification of the character corresponding to each item.

  In addition to the date notation format, the date dictionary 800 shown in FIG. 8 may be registered in the notation dictionary 309 when the item to be checked is a date. FIG. 8 is an explanatory diagram of the date dictionary 800 according to the first embodiment.

  The date dictionary 800 registers definitions of year, month, and day notations, and definitions of year, month, and day that are combinations thereof. The date dictionary 800 is described in BNF (Backus-Nur form) which is a notation of context free grammar.

  Next, screens displayed on the display unit 220 in this embodiment will be described with reference to FIGS. 9A to 9E. First, the character recognition unit 251 displays a place selection reception screen 900 shown in FIG. 9A and receives an input for selecting a place. Next, the character recognizing unit 251 displays a model number list display screen 910 (see FIG. 9B) that displays a list of model numbers of devices to be inspected at the place where the input is received on the place selection reception screen 900.

  When an image is input to the character recognizing unit 251, the character recognizing unit 251 recognizes the character of the model number, which is the first inspection item, from the input image, and the recognition result display screen 110 including the character recognition result of the model number (see FIG. 9C) is displayed. Next, the character recognizing unit 251 refers to the inspection scenario data 261, acquires the input location identifier and the inspection item of the equipment corresponding to the recognized model number, and includes the acquired inspection item list. A list display screen 920 (FIG. 9D) is displayed.

  Next, when a new image is input to the character recognizing unit 251, the character recognizing unit 251 recognizes the S / N character as the next inspection item from the input image, and the S / N character recognition result. A recognition result display screen including (see FIG. 9E) is displayed.

  Note that the character recognition unit 251 displays the recognition result display screen 110 every time a character of one item is recognized.

  FIG. 9A is an explanatory diagram of a place selection reception screen 900 according to the first embodiment.

  On the place selection reception screen 900, the name of the place identified by the identifier registered in the place 301 of the inspection scenario data 261 is displayed as options 901 and 902. The user operates the pointer 903 to select an option indicating a place where the inspection work is to be performed.

  FIG. 9B is an explanatory diagram of a model number list display screen 910 according to the first embodiment.

  When the character recognition unit 251 receives an input for selecting a place, the character recognition unit 251 refers to the inspection scenario data 261, acquires the model number registered in the model number 303 of the record in which the identifier of the place whose input is received in the place 301 is registered, A list of acquired model numbers is displayed as a model number list display screen 910.

  FIG. 9C is an explanatory diagram of a model number recognition result display screen 110 according to the first embodiment.

  When the image is input, the character recognition unit 251 first recognizes the character of the model number. The recognition result display screen 110 displays the character recognition result of the model number.

  The recognition result display screen 110 includes a captured image display area 120 and a character recognition result display area 130. The captured image display area 120 has the same configuration as that described with reference to FIG. The character recognition result display area 130 includes a recognized character string enlarged image display unit 131, a character recognition result display unit 132, a notation format display unit 931, an OK button 133, and a Retry button 134. The recognized character string enlarged image display unit 131, the character recognition result display unit 132, the OK button 133, and the Retry button 134 are the same as those described with reference to FIG.

  The notation format display unit 931 displays the notation format registered in the notation dictionary 309 used in the notation matching process shown in FIG. As a result, the user can check whether the character recognition result of the item matches the notation format. Note that the notation format display unit 931 may display a range such as numbers or letters, or may display a word registered in the word dictionary 308 used in the word matching process shown in FIG.

  In the recognition result display screen 110 illustrated in FIG. 9C, the character recognition result display unit 132 displays a cursor 932 that can be operated by the user. The recognized character string enlarged image display unit 131 displays a cursor 933 positioned on the character image corresponding to the character on which the cursor 932 is positioned.

  When the character recognition result displayed on the character recognition result display unit 132 is different from the character of the item to be inspected in the image, the user can move the cursor 932 to the character to be corrected and input the correct character. Since the cursor 933 is positioned on the image of the character where the cursor 932 is positioned, the user can immediately grasp the image corresponding to the character of the recognition result to be corrected.

  FIG. 9D is an explanatory diagram of an inspection item list display screen 920 according to the first embodiment.

  When recognizing the character of the model number, the character recognizing unit 521 refers to the inspection scenario data 261, registers the identifier of the location that has been input on the location selection reception screen 900 in the location 301, and the character recognized in the model number 303 An attribute registered in the attribute 305 of the registered record is acquired, and an inspection item list display screen 920 including the acquired attribute is displayed.

  On the inspection item list display screen 920, items to be inspected by the device to be inspected are displayed. Since the characters of the model number have already been recognized, the recognition result is displayed in the model number of the inspection item list display screen 920 shown in FIG. 9D.

  FIG. 9E is an explanatory diagram of the S / N recognition result display screen 110 according to the first embodiment.

  Next, when the character recognition of the model number is completed, the character recognition unit 251 receives a new image, recognizes the S / N character, and displays the S / N recognition result display screen 110. The configuration of the S / N recognition result display screen 110 is the same as the configuration of the model number recognition result display screen shown in FIG. 9C.

  The recognized character string enlarged image display unit 131 displays an enlarged image of the S / N character string, the character recognition result display unit 132 displays the recognition result of the S / N character, and the notation format display unit 931. The notation format registered in the notation dictionary 309 used in the S / N word matching process is displayed.

  When the OK button 133 is operated on the S / N recognition result display screen 110, the character recognition unit 251 displays an inspection item list display screen 920 shown in FIG. 9D. In this case, the S / N recognition result is displayed on the S / N of the inspection item list display screen 920.

  When a new image is input to the character recognition unit 251, the character of the next item (MAC address) is recognized, the MAC address recognition result display screen 110 is displayed, and the OK button 133 is operated, the inspection item A list display screen 920 is displayed. This process is repeated until the character of the last item is recognized.

  As described above, in this embodiment, the character recognition device recognizes characters included in the input image, and includes a plurality of groups including a plurality of items for recognizing characters and a recognition order of the plurality of items in the group. And a storage unit that stores a recognition dictionary (inspection scenario data 261) in which information used for character recognition of each item is registered, and a processor. The items registered in the recognition dictionary include a plurality of first items. It includes items (for example, device identification information such as model numbers) and a plurality of second items (items to be checked other than device identification information). The first item is the first item in the recognition order within the group, and the first item And at least one second item is associated with the processor, and the processor recognizes the character of the first item from the input image based on information used for recognizing the characters of the first item of the plurality of groups registered in the recognition dictionary. Shi Referencing the recognition order registered in the recognition dictionary, specifying the second item for recognizing the next character from the second item associated with the first item of the recognized character, and recognizing the character of the specified second item The information used for the second item is read from the recognition dictionary, and the character of the second item is recognized from the input image based on the read information used for the recognition of the character of the second item.

  As a result, the character of the item can be recognized using the information used for recognizing the character corresponding to the item, so that the character recognition accuracy can be improved, the user does not need to manually input each item, and the work efficiency is improved. Can be improved.

  In this embodiment, the information used for recognizing the first item character includes the first item notation format, and the processor recognizes the first item recognized character notation format as the first item. In the case of matching the notation format, the recognized character of the first item is taken as the recognition result of the first item. Thereby, the recognition accuracy of the first item can be improved. Since the next second item is determined by the recognition result of the first item, if the first item is erroneously recognized, the information used for recognizing the character of the second item is erroneously read. There is a need to improve accuracy.

  In this embodiment, the processor stores the recognized characters of the first item and the second item in the storage unit as the work log 262. Thereby, the previous character recognition result can be used for the next character recognition, and the recognition accuracy of each item can be improved.

  Further, in this embodiment, the processor corrects the input image based on the parameters, recognizes the character of the item to be recognized from the corrected image, and when the recognition of the character is successful, Parameters used for image correction are stored in the storage unit as the work log. As a result, the image can be corrected based on the parameter of the item that has been successfully recognized, so that the possibility that the corrected image is appropriate for character recognition increases, and the recognition accuracy can be improved.

  Further, in this embodiment, an image for recognizing the character of the first item is input, and the processor recognizes the character of the first item from the input image, and the recognized first item. An image for recognizing the second item character to be recognized next is input, and the processor recognizes the second item character from the input image and is associated with the recognized first item. An image for recognizing the last character of the second item is input, and the process is repeated until the character of the second item is recognized from the input image. For example, if an image in which an item to be recognized exists near the center is input, the recognition accuracy of the character of the item can be improved.

  Further, in this embodiment, the processor displays the recognized first item character on the display unit, and the recognized second item character in the recognition order registered in the recognition dictionary. To display. Thereby, the user can confirm the recognition result of the item according to the recognition order.

  Further, in this embodiment, the processor is registered in the recognition dictionary when displaying the recognized first item character on the display unit or when displaying the recognized second item character. The information used for the recognition is also displayed on the display unit. Thereby, the user can confirm the information used for recognition together with the recognition result of the item.

  Example 2 will be described with reference to FIGS.

  In the first embodiment, one image for recognizing a character of one item is captured by the camera and input to the character recognition unit 251, and character recognition processing is performed for each item. After the image for recognizing the model number character is input and the model number character is recognized, one image for recognizing the character of the item other than the model number is input, and all inspections other than the model number are input from the image The characters of items are recognized at once.

  FIG. 10 is a flowchart of character recognition processing according to the second embodiment. In FIG. 10, the same process as the character recognition process shown in FIG.

  The character recognition process of the present embodiment is different from the character recognition process of the first embodiment in the inspection item recognition step (530).

  In the inspection item recognition step of this embodiment, the inspection item recognition unit 253 executes an item collective recognition process (1001) and an item collective confirmation process (1002).

  In the item collective recognition process of step 1001, the characters of all the inspection items in the input image are recognized in order to recognize the characters of the items to be inspected. Details of the item batch confirmation processing will be described with reference to FIG.

  In the item batch confirmation process of step 1002, the recognition results of all the inspection items in the process of step 1001 are collectively confirmed. Details of the item batch confirmation processing will be described with reference to FIG.

  The character recognizing unit 251 determines whether or not the characters of all inspection items corresponding to the model number recognized in the process of step 504 have been recognized (1003).

  If it is determined in step 1003 that all the check item characters corresponding to the model number have been recognized, the process proceeds to step 510, and all devices installed in the place where the input is received in step 501 are processed. On the other hand, it is determined whether or not the processing of steps 520 and 530 has been executed.

  On the other hand, if it is determined in the process of step 1003 that the characters of all the check items corresponding to the model number are not recognized, whether or not the character string of the unrecognized item is included in the input image, It is conceivable that recognition of an input image containing a character string of an unrecognized item has failed. For this reason, the character recognizing unit 251 specifies an unrecognized item (1004), returns to the process of step 1001, receives an input of a new image, and executes the item recognition step again.

  FIG. 11 is a flowchart of item batch recognition processing according to the second embodiment.

  In the processing shown in FIG. 3 of the first embodiment, a new image is input for each inspection item, and character recognition preprocessing and character string recognition processing are executed on the input image. In the item collective recognition processing shown in FIG. 11, character recognition pre-processing and character string recognition processing are executed on one image input for character recognition of items other than the model number, and the one image other than the model number is processed. Recognize item characters. In FIG. 11, the same processes as those shown in FIG.

  If it is determined in step 601 that all items do not require character recognition (601: NO), the inspection item recognition unit 253 accepts input of data by the user and ends the item collective recognition processing.

  On the other hand, if it is determined in step 601 that any item requires character recognition (601: YES), the inspection item recognition unit 253 includes the location 301 among the records registered in the inspection scenario data 261. In step 501, the identifier of the place where the input was received is registered, and all records in which the model number recognized in step 504 is registered in the model number 303 are read (1101).

  Next, of the records registered in the work log 262, the inspection item recognizing unit 253 registers the identifier of the place that has received the input in the process of step 501 in the place 402, and recognizes the identifier of the model 404 in the process of step 504. All records with registered model numbers are read (1102).

  Next, the inspection item recognition unit 253 executes character recognition preprocessing (603). In the character recognition preprocessing of the present embodiment, the input image is corrected based on the parameters registered in the work log 262, and all character strings are extracted from the corrected image.

  Details of the character recognition pre-processing of the present embodiment will be described using the processing of steps 701 to 705 in FIG. In the processing in steps 702 to 704 of the first embodiment, various processes are executed on one image for recognizing a character of one item based on parameters included in the record of the item of the work log 262. The processing in steps 702 to 704 of the present embodiment is based on the parameters of each item of the device that performs the inspection work of the work log 262 for one image for recognizing the characters of all items other than the model number. Various processes. As a result, as many images as various types of processing are executed based on the parameters of each item are generated as many as the number of items of the device that performs the inspection work.

  In the processing of steps 702 to 704 of the present embodiment, the average value of each parameter of each item of the device that performs the inspection work of the work log 262 is calculated, and characters of all items other than the model number are recognized. Various processes may be executed on one image for the purpose based on the average value of each parameter. Alternatively, a parameter of one item may be selected, and various processes may be executed based on the parameter of one item for one image for recognizing characters of all items other than the model number.

  In the process of step 705 of the present embodiment, all character strings of the images for which the processes of steps 702 to 704 have been executed are extracted, and the process proceeds to the process of step 1103 shown in FIG.

  In the processing of step 1103, the inspection item recognition unit 253 registers the identifier of the location that has received the input in the processing of step 501 in the location 301 of the inspection scenario data 261, and the model number recognized in the processing of step 504 in the model number 303. The record of the item to be recognized next is read from the record in which is registered (1103).

  Next, the inspection item recognizing unit 253 registers the identifier of the place where the input has been received in the process of step 501 from the work log 262 in the place 402, and the model number recognized in the process of step 504 is registered in the model number 404. The record of the item to be recognized next is read from the record (1104).

  Next, the inspection item recognition unit 253 selects the first character string from the character strings extracted in step 603 (1105). Here, the first character string is the character string positioned at the top when the character string is extracted in order from the top of the image in the process of step 603. Note that the direction in which the character string is extracted in the process of step 603 is not limited from top to bottom, and may be extracted from the left to the right of the image, for example.

  If images corresponding to all items are generated in the character recognition pre-processing in step 603, the image is first extracted from the image corresponding to the record item in the inspection scenario data 261 read in the processing in step 1103. Select a string.

  Next, the inspection item recognition unit 253 executes a character string recognition process (604).

  The character string recognition process of the present embodiment will be described using the processes of steps 706 and 707 in FIG.

  In the process of step 706, the character recognition dictionary 711 is referred to and the character of the character string selected in the process of step 1105 is recognized.

  In the process of step 707, a word matching process and a notation matching process are executed on the character recognition result obtained in the process of step 706. In the word matching process, the check item recognition unit 253 determines whether the character recognition result obtained in step 706 matches the word registered in the word dictionary 308 of the item of the check scenario data 261 read in step 1103. Check whether or not. Further, in the notation matching process, the check item recognition unit 253 includes the notation format of the character recognition result in the process of step 706 and the notation registered in the notation dictionary 309 of the items of the inspection scenario data 261 read in the process of step 1103. Check whether the format matches.

  In the process of step 707 of this embodiment, a word matching score and a notation matching score are calculated.

  Next, the inspection item recognition unit 253 determines whether or not the character recognition in the process of Step 604 has been successful (1106). Specifically, the check item recognition unit 253 determines that the character recognition in the process of step 604 has been successful if at least one of the word matching score and the notation matching score is equal to or greater than a threshold value.

  If it is determined in step 1106 that the character recognition in step 604 has failed (1106: NO), the inspection item recognition unit 253 performs the process on all the character strings extracted in step 603. It is determined whether the process of step 604 has been executed (1107).

  If it is determined in step 1107 that the processing in step 604 has not been executed for all the character strings extracted in the processing in step 603 (1107: NO), the inspection item recognition unit 253 performs step 603. The next character string is selected from the character strings extracted in the process (1108), and the process returns to step 604.

  On the other hand, if it is determined in step 1107 that the process in step 604 has been executed for all the character strings extracted in step 603 (1107: YES), the inspection read in in step 1103 Since the character of the item of the scenario data 261 is not included in the input image or input, or the character recognition of the item has failed, the inspection item recognition unit 253 stores the item as an unrecognized item (1109). ), Selects the next item (1112), and returns to the processing of step 1103. In step 1103, the record of the next item selected from the inspection scenario data 261 is read.

  If it is determined in step 1106 that the character recognition in step 604 has been successful (1106: YES), the inspection item recognition unit 253 stores the character recognition result in step 604 (1110). Then, it is determined whether or not the process of step 604 has been executed for all items of the equipment to be inspected (1111).

  If it is determined in step 1111 that the process in step 604 has been executed for all items of the equipment to be inspected (1111: YES), the item collective recognition process is terminated.

  On the other hand, if it is determined in step 1111 that the process of step 604 has not been executed for all items of the equipment that performs the inspection work (1111: NO), the process proceeds to step 1112 to check the items to be checked. The recognition unit 253 selects the next item and returns to the processing of step 1103.

  FIG. 12 is a flowchart of item batch confirmation processing according to the second embodiment.

  First, the inspection item recognition unit 253 selects the first item recognized in the item collective recognition process (1201).

  Next, the inspection item recognition unit 253 displays the recognition result display screen 110 including the character recognition result of the item selected in step 1201, and presents the character recognition result of the item to the user (1202). In the process of step 1202, the user looks at the recognition result display screen 110 and determines whether or not the character recognition result of the item selected in the process of step 1201 is correct. If the user determines that the character recognition result is incorrect, the user causes the inspection item recognition unit 253 to execute the character batch recognition process again, or inputs correction contents to the character recognition device 200.

  Next, the inspection item recognition unit 253 registers the character recognition result selected in the process of step 1201 or the character recognition result corrected by the user and the parameters used in the process of step 603 in the work log 262 (1203). .

  Next, the inspection item recognition unit 253 determines whether or not the processing of steps 1202 and 1203 has been performed on all items recognized in the item collective recognition processing (1204).

  If it is determined in step 1204 that the processing in steps 1202 and 1203 has not been executed for all items recognized in the item collective recognition processing (1204: NO), the inspection item recognition unit 253 performs inspection collective recognition. The next item recognized in the process is selected (1205), and the process returns to step 1202.

  On the other hand, when it is determined in the process of step 1204 that the processes of steps 1202 and 1203 have been executed for all items recognized in the item collective recognition process (1204: YES), the inspection item recognition unit 253 It is determined whether or not an unrecognized item is stored in the process (1206).

  If it is determined in step 1206 that an unrecognized item is stored in step 1109 (1206: YES), inspection item recognition unit 253 lists unrecognized items stored in step 1109. The unrecognized item display screen 1400 including "1207" is displayed (1207), and the item batch confirmation process is terminated. Details of the unrecognized item display screen 1400 will be described with reference to FIG.

  On the other hand, in the process of step 1206, when it is determined that the unrecognized item is not stored in the process of step 1109 (1206: NO), the inspection item recognition unit 253 ends the item batch confirmation process.

  Next, screens displayed in the present embodiment will be described with reference to FIGS. 13A to 14.

  First, the inspection item recognition unit 253 displays the model number recognition result display screen 110 before the execution of the step 1201 of the item collective confirmation processing shown in FIG. FIG. 13A is an explanatory diagram of a model number recognition result display screen 110 according to the second embodiment.

  Similar to the recognition result display screen 110 described in FIG. 1, the model number recognition result display screen 110 shown in FIG. 13A includes a captured image display area 120 and a character recognition result display area 130.

  In the captured image display area 120, an image input in the process of step 701 of the character recognition pre-processing of the item batch recognition process is displayed. Further, in the image displayed in the captured image display area 120, the image of the character string in which the character of the inspection item is recognized is displayed surrounded by the rectangles 1301 and 1302.

  The character recognition result display area 130 includes a model number recognition result display unit 1311 and an inspection item display unit 1312. The model number recognition result display unit 1311 displays the characters of the model number recognized by the model number verification 504. The inspection item display unit 1312 displays items to be inspected by the recognized model number device in accordance with the inspection order.

  When executing the item batch confirmation process on the image displayed in the captured image display area 120, the user operates the OK button 133. When a character string to be inspected is not included in the image displayed in the captured image display area 120 and a new image is input, the user operates the Retry button 134.

  Next, in the process of step 1201 of the item batch confirmation process shown in FIG. 12, the item (S / N) to be inspected first is selected, and in the process of step 1202, the recognition result display screen 110 of the item (S / N) is selected. Is displayed. FIG. 13B is an explanatory diagram of an item (S / N) recognition result display screen 110 according to the second embodiment.

  In the image displayed in the captured image display area 120 of the item (S / N) recognition result display screen 110, a rectangle 1301 surrounding the image of the S / N character string that is the item of the recognition result is highlighted. On the other hand, the rectangle 1302 surrounding the MAC character string image which is not an item of the recognition result is not highlighted.

  The character recognition result display area 130 is the same as that in FIG.

  When the user determines that the recognition result is correct on the item (S / N) recognition result display screen 110, or when the correction of the recognition result is completed, the user operates the OK button 133. When the user operates the OK button 133, the inspection item recognition unit 253 displays the recognition result display screen 110 (see FIG. 13C) for the next item (MAC). On the other hand, when the user notices that the character string of the item to be checked is not included in the image displayed in the captured image display area 120, the user operates the Retry button 134. In this case, the inspection item recognition unit 253 receives an input of a new image and executes the item collective recognition process again.

  FIG. 13C is an explanatory diagram of an item (MAC) recognition result display screen 110 according to the second embodiment.

  In the image displayed in the captured image display area 120 of the item (MAC) recognition result display screen 110, a rectangle 1302 surrounding the MAC character string image that is the item of the recognition result is highlighted. On the other hand, a rectangle 1301 surrounding an image of an S / N character string that is not a recognition result item is not highlighted.

  The character recognition result display area 130 is the same as that in FIG.

  FIG. 14 is an explanatory diagram of an unrecognized item display screen 1400 according to the second embodiment.

  The unrecognized item display screen 1400 displays a list of items that are not recognized in the item collective recognition process among the items to be inspected of the device that performs the inspection work.

  As described above, in this embodiment, an image for recognizing the first item character is input, and the processor recognizes the first item character from the input image, and the recognized first item. An image for recognizing the character of the second item associated with is input, and the processor recognizes the character of the second item associated with the recognized first item from the input image. As a result, since the character of the first item is recognized from the image for the first item, which is an item necessary for specifying the second item, the second item is one item while improving the recognition accuracy of the first item. Since it recognizes collectively from an image, it becomes unnecessary to input an image for every 2nd item, and a user's working efficiency can be improved.

  A third embodiment will be described with reference to FIGS. 15 to 16B.

  In Example 2, after the image for recognizing the model number character is input and the model number character is recognized, one image is input, and the characters of all the inspection items other than the model number are collectively collected from the image. Although recognized, in this embodiment, characters of all items including the model number are collectively recognized from one input image.

  FIG. 15 is a flowchart of character recognition processing according to the third embodiment. In FIG. 15, the same process as the character recognition process shown in FIG. 5 of the first embodiment and FIG. 10 of the second embodiment is given the same reference numeral, and the description is omitted.

  In the character recognition process of this embodiment, the process of step 504 (model number verification process) shown in FIG. 10 is not executed, and all item batch recognition is performed instead of the process of step 1001 (item batch recognition process) shown in FIG. The process is executed (1501), and the all item collective confirmation process is executed (1502) instead of the process of step 1002 (item collective confirmation process) shown in FIG.

  In the all item collective recognition process executed in step 1501, the character recognition unit 251 extracts a character string from the input image, and recognizes characters of all items including the model number from the extracted character string.

  The all item collective recognition process can be basically realized by the same process as the item collective recognition process. However, the all item collective recognition process is different from the item collective recognition process shown in FIG. 10 in that characters of all items including the model number are recognized.

  In the all item collective recognition process, it is necessary to first recognize the character of the model number from the input image. For this reason, in the process of step 1103 before the characters of the model number are recognized, the character recognition unit 251 records the model number records of all the devices installed in the place where the input is received in the process of step 501 as the inspection scenario data 261. The character of the model number is recognized in the process of step 604. Alternatively, model number records are read one by one in the order of the device number 302 of the inspection scenario data 261, and the characters of the model number are recognized in the process of step 604. Also in the process of step 1104, the record of the model number of the device installed at the place where the input is received is read from the work log 262 as in the process of step 1103. The all item collective recognition process after the character of the model number is recognized is the same as the item collective recognition process.

  The all item collective confirmation process can be basically realized by the same process as the item collective recognition confirmation process. The all item batch check process is different from the item batch check process in that a recognition result display screen 110 showing the character recognition result of the model number is also displayed.

  Next, a screen displayed in this embodiment will be described with reference to FIGS. 16A and 16B.

  First, the character recognition unit 251 displays a recognition result display screen 110 that displays a list of inspection items before displaying the recognition result display screen 110 (see FIG. 16B) of the first item (model number) in the all item batch confirmation process. indicate. FIG. 16A is an explanatory diagram of a recognition result display screen 110 that displays a list of inspection items according to the third embodiment.

  The recognition result display screen 110 illustrated in FIG. 16A includes a captured image display area 120 and a character recognition result display area 130.

  In the captured image display area 120, an image input in the batch recognition process for all items is displayed. In the image displayed in the photographed image display area 120, an image of a character string recognized as a model number is surrounded and displayed by a rectangle 1601, and characters of an image inspection item of a character string recognized as an inspection item other than the model number are displayed. An image of a character string in which is recognized is surrounded by rectangles 1301 and 1302 and displayed.

  The rectangle 1601 surrounding the character string of the model number and the rectangles 1301 and 1302 surrounding the character string of the inspection item are displayed so that the user can specify whether they are the character string of the model number or the character string of the inspection item. For example, the rectangle 1601 is displayed as a solid line, and the rectangles 1301 and 1302 are displayed as dotted lines.

  The character recognition result display area 130 includes an inspection item display unit 1312. The inspection item display unit 1312 displays items including the model number to be inspected in the inspection order.

  When the OK button 133 is operated, the character recognition unit 251 displays the recognition result display screen 110 of the model number that is the first item. FIG. 16B is an explanatory diagram of a model number recognition result display screen 110 according to the third embodiment. The model number recognition result display screen 110 shown in FIG. 16B is the same as the item (S / N) recognition result display screen 110 shown in FIG. 13B except that the model number recognition result is displayed, and the same configuration is the same. Reference numerals are assigned and description is omitted.

  In the image displayed in the captured image display area 120, the character string rectangle 1601 of the model number whose recognition result is displayed in the character recognition result display area 130 is highlighted with a solid line, for example. Note that the rectangle 1301 surrounding the S / N character string and the rectangle 1302 surrounding the MAC character string are displayed as dotted lines, and the distance between the dotted lines of the rectangle 1301 where the recognition result is displayed next is expressed as By increasing the distance between the dotted points of the rectangle 1302, the rectangle surrounding the character string of the item for which the recognition result is displayed next and the rectangle surrounding the character string of the other item can be easily grasped. To do.

  When the OK button 133 is operated, the character recognition unit 251 displays a recognition result display screen 110 for the next item (MAC).

  In the present embodiment, the processor recognizes the character of the first item from one input image, and from all the second items associated with the recognized first item from the input image. Recognize characters at once. As a result, the characters of all items can be recognized at once by inputting only one image, so that the user's work efficiency can be improved.

  In this embodiment, the character recognition unit 251 performs an inspection operation on a certain device, and when the character of the item of the device is recognized, the character recognition unit 251 displays the relative arrangement relationship (layout) of the item on which the character is recognized as a work log. As 262. The arrangement relationship may be, for example, a positional relationship such that the S / N is located below the model number and the MAC is located below the S / N, or may be the distance between the centroids of the items. . The positional relationship is not limited to the vertical direction, and may include a horizontal positional relationship.

  When the inspection work is performed with the same equipment, the character recognition unit 251 is input with reference to the layout of the record of the item recognized in the work log 262 and an area for extracting the character string of the item that recognizes the character. Identify from the image. Then, the character recognition unit 251 extracts a character string from the identified area, and recognizes the character in the extracted character string. As described above, since the character string is extracted from the extracted position of the character string specified with reference to the layout of the work log 262, it is not necessary to extract the character string from all the positions of the input image. 200 processing loads can be reduced. This embodiment can be applied to Embodiments 2 and 2 in which processing for recognizing a plurality of items at once is executed.

  A character recognition process according to this embodiment will be described with reference to the second embodiment.

  First, in the process of step 1203 of the item batch confirmation process shown in FIG. 10, when registering the character recognition result of the item displayed on the recognition result display screen 110 of step 1202 and the parameters for correcting the image in the work log 262. In addition, coordinates indicating the position of the character string in which the character of the item is recognized are also registered in the work log 262. For example, the coordinates registered in the work log 262 are the coordinates of the center of gravity of the area where the character string of the input image is extracted. Then, the character recognition unit 251 calculates the direction and distance from the character string coordinate of the item immediately before the item number of the item to the character string coordinate of the item, and calculates the calculated direction and distance of the item. Register in the work log 262. As described above, in the item batch confirmation process of this embodiment, the direction and distance from the coordinates of the character string of the previous item are registered as a layout in the work log 262 of the item to be confirmed.

  Next, the item collective recognition process using the layout registered in the work log 262 will be described with reference to FIG. In the second embodiment, the character string is extracted from the image input in the character recognition pre-processing in step 603, but in this embodiment, the character string is not extracted in the character recognition pre-processing in step 603. In step 1105, a character string is extracted, and the extracted character string is selected.

  A character string extraction process executed in the process of step 1105 of this embodiment will be described. The character recognizing unit 251 acquires a layout from the work log 262 of the item for recognizing the character read in the process of step 1104. Then, in the processing of step 1105, the character recognition unit 251 determines the character string of the area specified by the direction and distance included in the acquired layout from the position of the character string of the item immediately preceding the item for recognizing the character. The extracted character string is selected, and the character string recognition process in step 604 is executed.

  As described above, in this embodiment, since the character string is extracted from the area limited for each item without extracting all the character strings of the input image, the processing load of the character recognition device 200 can be reduced. In addition, since the character string is extracted from the region where the character string of the item for recognizing the character is highly likely to be located, and the character of the extracted character string is recognized, the recognition accuracy can be improved.

  Note that when the same equipment is inspected multiple times, the angle of view of the camera may vary from inspection to inspection, but the character string can be accurately corrected by correcting the difference in the angle of view in the character recognition pre-processing. Can be extracted. Further, by statistically analyzing the layout of items for each inspection operation and registering the analysis result in the operation log 262, the difference in the angle of view can be absorbed.

  Further, the layout of the work log 262 may include the coordinates of character strings around the character string of the item. In this case, when extracting the character string of the item for recognizing the character, the character recognizing unit 251 does not extract the character string near the coordinates of the surrounding character string that is not the recognition target. Since the character string around the item for character recognition is not extracted, the number of candidate character strings can be reduced, and the recognition accuracy can be improved.

  In the first to third embodiments, the image input to the character recognition unit 251 is described as a still image. However, the image input to the character recognition unit 251 may be a moving image. The user may shoot a moving image such as a label of a device on which an inspection operation is performed, the moving image may be input to the character recognition unit 251 in real time, and the character recognition unit 251 may execute the character recognition processing in real time. Note that the character recognition unit 251 may end the shooting of the moving image and allow the user to check the recognition result when the character recognition of all items of a certain device is completed.

  In the first to third embodiments, the character recognition unit 251 recognizes characters from one image, but may recognize characters from a plurality of images. For example, a moving image is input to the character recognition unit 251, a character is recognized from a plurality of frame images included in the input moving image, and if a predetermined number or more of the same number of frame images have the same recognition result, the recognition result is stored. Thereby, recognition accuracy can be improved.

  The characters recognized in the first to third embodiments are character strings to be inspected written around the name of the item, but recognize the character string of the item name and the character string to be inspected of the item. May be. This is effective when a character string with the same notation exists in one image. For example, this is the case where the image input to the character recognition unit 251 includes the date of the next scheduled inspection date and the date of the previous inspection date. For example, from the label 141 shown in FIG. 1, the combination of the item name “S / N” and the character string “1710401161” to be inspected, and the item name “MAC” and the character string “C08ADE14CB60 to be inspected”. ”Is extracted.

  Further, it is possible to add, delete, and replace other configurations for a part of the configuration of each embodiment. Each of the above-described configurations, functions, processing units, processing means, and the like may be realized by hardware by designing a part or all of them with, for example, an integrated circuit.

  Each of the above-described configurations, functions, and the like may be realized by software by interpreting and executing a program that realizes each function by the processor.

  Information such as programs, tables, and files that realize each function is stored in memory, a hard disk, a recording device such as an SSD (Solid State Drive), an IC (Integrated Circuit) card, an SD card, a DVD (Digital Versatile Disc), etc. Can be placed on any recording medium.

  Further, the control lines and information lines indicate what is considered necessary for the explanation, and not all the control lines and information lines on the product are necessarily shown. Actually, it may be considered that almost all the components are connected to each other.

200 Character Recognition Device 210 Operation Unit 220 Display Unit 230 Image Input Unit 250 Processor 251 Character Recognition Unit 252 Device Specific Information Recognition Unit 253 Inspection Item Recognition Unit 260 Memory 261 Inspection Scenario Data 262 Work Log 270 Auxiliary Storage Device

Claims (12)

A character recognition device for recognizing characters included in an input image,
A recognition dictionary in which a plurality of groups including a plurality of items for recognizing characters, a recognition order of a plurality of items included in the group in the group, and information used for recognizing characters of each item are stored. A storage unit to
And a processor,
Items registered in the recognition dictionary include a plurality of first items and a plurality of second items,
The first item is the first item in the recognition order in the group, and at least one second item is associated with the first item;
The processor is
Recognizing characters of the first item from an input image based on information used for recognizing characters of the first item of a plurality of groups registered in the recognition dictionary;
Referring to the recognition order registered in the recognition dictionary, specifying a second item for recognizing a character next from a second item associated with the first item of the recognized character;
Read information used for recognizing the character of the specified second item from the recognition dictionary,
A character recognition device that recognizes the character of the second item from the input image based on the information used for recognizing the character of the read second item. The character recognition device according to claim 1,
The information used for recognizing the character of the first item includes the notation format of the first item,
When the notation format of the recognized character of the first item matches the notation format of the first item, the processor sets the recognized character of the first item as a recognition result of the first item. Character recognition device. The character recognition device according to claim 1,
The processor stores the recognized characters of the first item and the second item in the storage unit as a work log. The character recognition device according to claim 3,
The processor is
Correct the input image based on the parameters, recognize the character of the item to be recognized from the corrected image,
The character recognition device, wherein when the character recognition is successful, the storage unit stores the parameters used for correcting the image as the work log. The character recognition device according to claim 3,
The processor is
Extract a character string from the input image,
Recognizing characters based on information used for recognizing characters of items to be recognized from the extracted character strings,
When the recognition of the character is successful, the arrangement relationship between the item that recognized the character and the item whose recognition order is one before the item is registered in the work log,
When recognizing the character of the item, based on the arrangement relationship registered in the work log, specify the area to extract the character string from the input image,
A character recognition apparatus, wherein a character string is extracted from the specified area. The character recognition device according to claim 1,
An image for recognizing the character of the first item is input, and the processor recognizes the character of the first item from the input image,
An image for recognizing a second item character recognized next to the recognized first item is input, and the processor recognizes the second item character from the input image,
An image for recognizing the character of the last second item associated with the recognized first item is input, and the process is repeated until the character of the second item is recognized from the input image. Character recognition device. The character recognition device according to claim 1,
An image for recognizing the character of the first item is input, and the processor recognizes the character of the first item from the input image,
An image for recognizing the character of the second item associated with the recognized first item is input, and the processor selects all the second items associated with the recognized first item from the input image. 2. A character recognition apparatus characterized by recognizing two items of characters collectively. The character recognition device according to claim 1,
The processor is
Recognizing the characters of the first item from one input image,
A character recognition apparatus that collectively recognizes characters of all second items associated with the recognized first item from the input image. The character recognition device according to claim 1,
The processor is
Displaying the character of the recognized first item on the display unit;
The character recognition apparatus, wherein the recognized second item character is displayed on the display unit in a recognition order registered in the recognition dictionary. The character recognition device according to claim 9,
The processor, when displaying the recognized first item of characters on the display unit, or when displaying the recognized second item of characters on the display unit, the processor registered in the recognition dictionary The character recognition apparatus characterized by also displaying the information used for recognition on the said display part. A character recognition method in a character recognition device for recognizing characters included in an input image,
The character recognition device includes a storage unit and a processor,
Registered in the storage unit are a plurality of groups including a plurality of items for recognizing characters, a recognition order of a plurality of items included in the group in the group, and information used for character recognition of each item The recognized recognition dictionary is stored,
Items registered in the recognition dictionary include a plurality of first items and a plurality of second items,
The first item is the first item in the recognition order in the group, and at least one second item is associated with the first item;
The character recognition method is:
The processor recognizes the character of the first item from the input image based on information used for recognizing the character of the first item of a plurality of groups registered in the recognition dictionary;
The processor refers to the recognition order registered in the recognition dictionary, specifies a second item for recognizing a character next from a second item associated with the first item of the recognized character,
The processor reads information used for recognizing the character of the specified second item from the recognition dictionary,
The character recognition method characterized by recognizing the character of the said 2nd item from the input image based on the information used for the recognition of the character of the said read 2nd item of the said processor. A program for causing a processor included in an apparatus to execute processing for recognizing characters included in an input image,
The apparatus includes a storage unit and the processor,
Registered in the storage unit are a plurality of groups including a plurality of items for recognizing characters, a recognition order of a plurality of items included in the group in the group, and information used for character recognition of each item The recognized recognition dictionary is stored,
Items registered in the recognition dictionary include a plurality of first items and a plurality of second items,
The first item is the first item in the recognition order in the group, and at least one second item is associated with the first item;
In the process,
The processor recognizes the character of the first item from the input image based on information used for recognizing the character of the first item of a plurality of groups registered in the recognition dictionary;
The processor refers to the recognition order registered in the recognition dictionary, specifies a second item for recognizing a character next from a second item associated with the first item of the recognized character,
The processor reads information used for recognizing the character of the specified second item from the recognition dictionary,
The program for recognizing the character of the second item from the input image based on the information used for recognizing the character of the read second item.