JPH1034089A - Video coding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a video coding device which enables an operator to efficiently and rapidly input user codes with a less fatiguing operation and to embody rapid processing. SOLUTION: This video coding device takes in the entire part image on postal items stated with the user codes consisting of postal code numbers and address display numbers and addressee address information and the information on the results of the recognition for the same from a reading section 2, displays the same on a display section 32, inputs the user codes on the postal items by a keyboard 31 in accordance with the displayed images and transmits the user codes to a bar code printing section. The operator inputs town names and bulk posting receiver names by voices from a voice input section 34 to the postal items, such as those not stated with the user codes, those stated with the incomplete user codes and those having a difficulty in reading of the user codes. These voices are then recognized in a voice recognition section 30 and the candidates for voice recognition having a high probability are displayed in the display section 32. The operator inputs the user codes by selecting and determining the user codes out of the displayed candidates for voice recognition.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention reads, for example, a user code consisting of at least a postal code and an address display number, which is described in a user code designation entry area set in advance on a postal matter. In a mail processing device that prints barcode information corresponding to the read user code on a mail and sorts mail based on the barcode information, the mail that cannot read the user code is read. The operator inputs the user code while viewing the displayed whole image on the mail, converts it into the corresponding bar code information, and prints it on the mail, so that the The present invention relates to a video coding device for sorting mail.

[0002]

2. Description of the Related Art Recently, mail processing apparatuses using a video coding apparatus have been developed. This type of mail processing apparatus reads, for example, a user code (for example, consisting of a zip code and an address display number) indicating an address written on a mail by an optical character reading apparatus, and reads it. By converting the barcode information into barcode information and printing it on the mail, the mail is sorted according to the barcode information for each delivery area.

[0003] For mail items for which the user code could not be read, the entire image is displayed on the display unit of the video coding apparatus, and the operator inputs the user code while visually observing the display screen. By converting it to the corresponding barcode information and printing it on mail,
The mail is sorted according to the barcode information. This makes it possible to compensate for the limitations of the machine classification by the optical character reader.

[0004] However, in the conventional video coding apparatus, the operator reads the user codes of all rejected mail items because the user codes cannot be read, and the image information (video image) is read by the operator. It was coded by key input.

[0005]

As described above, in the conventional video coding apparatus, the user codes of all rejected mails are read by an operator by reading the image information, and are all coded by key input. I was For this reason, the number of keystrokes increases in proportion to the number of input codes, so that there has been a problem that the input speed does not increase and the accuracy decreases.

[0006] Further, actual postal items include those in which the user code is not described or incomplete, those which are difficult to read, and those in which the wrong user code is described. For such mails, it is difficult to improve the input speed, and the input rhythm of continuous input is disrupted, thereby reducing the input efficiency of the entire system.

Therefore, the present invention provides a video coding apparatus which enables an operator to input a user code efficiently and in a short time with less fatigue, to improve accuracy and labor saving, and to realize quick processing. The purpose is to:

[0008]

SUMMARY OF THE INVENTION A video coding apparatus according to the present invention comprises: an entire image on a postal matter in which a user code including at least a postal code and an address display number and address information are described; The system retrieves and displays the recognition result information for the entire image of the postal matter, and inputs the user code on the postal matter based on the displayed image. An entire image, first input means for inputting a candidate character of a user code recognized by the recognition means for the mail, and an entire image of the mail input by the first input means; And first display means for displaying candidate characters of the user code and displaying a user code input area for inputting the user code, respectively; While viewing the whole image of the mailpiece,
A second input means for inputting a user code in a user code input area which is displayed at the same time, or a voice recognition target type information, and an entire image of the mail displayed on the first display means Voice input means for inputting a voice recognition target name based on a voice, voice recognition means for recognizing voice information input by the voice input means, recognition target vocabulary reading information for each voice recognition target type, and corresponding A first storage unit that stores a recognition target registration code, a voice recognition target type information input by the second input unit, and the first storage unit based on a recognition result of the voice recognition unit. Searching means for extracting at least one candidate for the recognition target registration code, and a second storage means for storing recognition target display information corresponding to the candidate for the recognition target registration code. Storage means and at least one or more recognition target display information corresponding to a candidate for a recognition target registration code by searching the second storage means based on a candidate for a recognition target registration code output from the search means. And a recognition target candidate selecting means for selecting and determining a recognition target candidate from the recognition target display information displayed on the second display means.

According to the present invention, for example, a mail in which a user code is not described, a user code is incomplete, a user code is difficult to read, and an incorrect user code is described, for example. By inputting the voice recognition target name such as the name of the town, the name of the major recipient, etc.
Speech recognition is performed, and a speech recognition target candidate with high accuracy is displayed. The operator can accurately input the user code only by selecting and determining from the displayed voice recognition target candidates. Therefore, the operator can input the user code efficiently in a short period of time with less fatigue, improving accuracy and labor saving, and realizing quick processing.

The first storage means for storing recognition target vocabulary reading information for performing a search based on the result of speech recognition and a recognition target registration code corresponding to the recognition target vocabulary reading information includes: It is possible to have the target vocabulary reading information, and the second display means displays the code as a single recognition target display information even if there are a plurality of candidates for the same recognition target registration code, so that the code by voice recognition is displayed. Accuracy and speed of conversion are improved.

[0011] Also, regarding the display of the voice recognition target candidates, the operator can promptly and accurately recognize the voice recognition by displaying the voice recognition target candidate having the highest recognition accuracy separately from the other voice recognition target candidates. It becomes possible to select target candidates.

The code corresponding to the voice recognition target candidate having the highest recognition accuracy is displayed in the user code input area from the beginning, and is temporarily selected during the selection of the voice recognition target candidate. An efficient system can be constructed by sequentially displaying codes corresponding to voice recognition target candidates in a user code input area.

Further, the speech recognition target type information including the presence / absence information of the second and subsequent voice input processes is input to the postal matter to be voice-recognized. If yes, since the recognition target vocabulary reading information extracted as the speech recognition target candidate in the previous speech input is excluded and the first storage means is searched, the recognition target registration code candidate is obtained. , Efficient,
A highly accurate system can be realized.

[0014]

Embodiments of the present invention will be described below with reference to the drawings. 1 and 2 show a postal matter according to the present embodiment. FIG. 1 shows an example of a handwritten postal matter (postcard) P, in which a user code (division of municipalities + chome numbers) in place of a conventional postal code or address information is a new postal code (city). It is composed of a section up to the ward, town and village, described in a red frame) and an address display number (a section up to a street number and a building room number, described in an address information area). That is, for example, the upper three digits and the lower four
A new zip code designation area 61 in which a new zip code consisting of a total of seven digits (division up to municipalities, described in a red frame) is set in advance. Reference numeral 62 denotes an address information area in which address information including an address display number (division up to a street number and a ridge room number) is handwritten.

FIG. 2 shows an example of a print type postal matter (postcard) P. On the surface of the postal matter P, a new postal code designation area 61 is set in advance similarly to a handwritten postal matter. However, in this case, the new zip code is stored in the new zip code designation area 61 similarly to the address display number.
Are often described in print type together with the address information in the address information area 62 as shown in the figure.

FIG. 3 schematically shows a configuration of a mail processing apparatus to which the video coding apparatus according to the present embodiment is applied. That is, a large number of postal items P configured as described above are set in the supply unit 1 collectively. The supply unit 1 supplies the set mail items P to the reading unit 2 one by one.

The reading unit 2 reads (recognizes) the user code written on the mail P based on the image by taking in the whole image on the mail P supplied. When the user code is read, the user code is output to the barcode printing unit 3. At this time, regardless of whether or not reading is possible, the reading unit 2 prints a serial number (hereinafter abbreviated as an ID number) corresponding to each mail P on the barcode printing unit 3.
Output to

The postal matter P whose user code has been read by the reading unit 2 is formed by a long transport path.
The postal matter P is conveyed to the barcode printing unit 3 via a delay path 4 which delays the conveyance of the postal matter P to the subsequent stage. The barcode printing unit 3 converts the user code from the reading unit 2 into corresponding barcode information, and prints on the mail P while confirming the mail P associated with the ID number.

The postal matter P on which the barcode information is printed by the barcode printing unit 3 is sent to the barcode reading unit 5.
The barcode reading unit 5 reads and decodes the barcode information printed on the mail P, and sends the reading result (user code) to the sorting unit 6 together with the mail P. Division 6
The postal matter P is classified for each delivery area based on the result of reading from the barcode reading unit 5.

For the mail P whose user code cannot be read by the reading section 2, the corresponding mail P
Image, ID number, attribute information (classification other than postcard / postcard, classification of handwritten mail / printed mail, character line direction judgment result, etc.), user code recognition result, and user code description area Is sent to the video coding device 7 and a reject signal is sent to the barcode printing unit 3.

In the video coding device 7, the whole image of the mail P from the reading unit 2 is displayed on the display unit, as will be described in detail later. The operator inputs (or corrects) the user code by input operation with a keyboard, a mouse, or the like while visually checking the display image, and after the user code is checked for correctness, the bar code printing unit. Sent to 3.

On the other hand, the postal matter P whose user code could not be read is conveyed to the bar code printing unit 3 via the delay path 4. The bar code printing unit 3 converts the user code from the video coding device 7 into bar code information based on a reject signal for the postal matter P from which the user code output from the reading unit 2 cannot be read at this time. It is printed on the mail P correlated by the number.

[0023] The subsequent steps are the same as those of the postal matter P from which the user code has been read, and a description thereof will be omitted. FIG. 4 shows the configuration of the reading unit 2. That is, the photoelectric conversion unit 11 irradiates light on the mail P supplied from the supply unit 1 from a light source, and converts the reflected light into C light.
After being converted into an electric signal by a CD type line sensor or the like, the whole image of the mail P is captured by digitizing.

The entire image of the postal matter P captured by the photoelectric conversion unit 11 is temporarily stored in the image memory 12, and
It is sent to the area detection unit 13. The area detection unit 13 detects a new postcode code designation area 61 and an address information area 62 on the postal matter P from the input image of the postal matter P, and re-reads the information from the user code recognition unit 18 described later. When the read command signal is input, the user code description area other than the new postal code code specification description area 61 is detected again from the input image of the postal matter P.

The output of the area detection unit 13 is
4 The character line detection unit 14 detects a character line by obtaining, for example, a horizontal projection pattern from the image in the detected area.

The output of the character line detector 14 is
Sent to 5. The character detection unit 15 detects a character for each character by obtaining, for example, a projection pattern in the vertical direction from the detected character line.

The output of the character detecting unit 15 is
Sent to The character recognizing unit 16 performs a recognition process on the detected character by referring to the character dictionary 17. The recognition result of the character recognition unit 16 is sent to a user code recognition unit 18 as recognition means. The user code recognition unit 18 performs a user code recognition process by performing a knowledge process using the user code dictionary 19 based on the recognition result of the character recognition unit 16. If the user code can be recognized in this recognition processing, the user code is output to the barcode printing unit 3.

In the above recognition process, if the user code cannot be recognized, a re-read command signal is sent to the area detecting unit 13.
, The user code is recognized again. In this recognition process, if the user code cannot be recognized, it is finally determined to be unrecognizable, and an ID number for associating with the postal matter P and a reject indicating that the user code has not been recognized. The signal is output to the video coding device 7.

When the user code recognizing unit 18 cannot recognize the user code, the user code candidate character information and the postal matter P attribute information (other than postcard / postcard) are used as recognition result information at this time. Classification, classification of handwritten mail / printed print mail, character line direction determination result, etc.), coordinate information of the new zip code designation area 61 detected by the area detection unit 13, and new zip code designation area 61
It outputs to the video coding device 7 coordinate information of a user code description area other than the above (that is, user code description candidate area information) and the like.

Here, the classification of the handwritten mail / printed print mail is, for example, when the new zip code is described in the new zip code designation area 61, the mail is judged as a handwritten mail, and the new mail is determined. If the number is not described, it is classified by judging it as a print type mail.

The user code recognizing section 18 outputs an ID number corresponding to the mail P to the bar code printing section 3 regardless of whether the user code can be recognized. Further, if the user code cannot be recognized by the user code recognition unit 18, the entire image of the corresponding mail P in the image memory 12 is output to the video coding device 7.

Next, the video coding device 7 will be described in detail. FIG. 5 schematically shows the configuration of the video coding device 7. That is, a CPU (Central Processing Unit) 21 as control means for controlling the entire apparatus includes a ROM (Read Only Memory) 22 storing fixed information such as a control program of the CPU 21, a working memory. RAM (random access memory) 2 used as
3, a first storage unit 24 as a storage unit, a second storage unit 25 as a storage unit, an input interface 26 as an input unit, a keyboard control unit 27, a display control unit 28,
The print control unit 29 and the voice recognition unit 30 are connected to each other.

The input unit 26 includes the reading unit 2
However, the keyboard control unit 27 has a keyboard 31 as input means, and the display control unit 28 has a CR as display means.
A display unit 32 such as a T display and a printer 33 are connected to the print control unit 29, respectively.

The first storage section 24 temporarily stores image data and secures a specific area shown in FIG. 6, which will be described later. For example, a semiconductor memory such as a RAM or an EEPROM is used.

The second storage unit 25 is formed of, for example, a large-capacity hard disk device. When a user code is input by the keyboard 31, the second mail unit 25 stores the input mail P to support the input operation. An overall image, a recognition result of the reading unit 2, an input user code, an address database for checking the input user code, and the like are stored.

The input interface 26 is constituted by, for example, a so-called SCSI interface. The whole image of the postal matter P whose user code could not be determined by the reading unit 2, As recognition results, attribute information (classification other than postcard / postcard, classification of handwritten mail / printed mail, character line direction judgment result, etc.), candidate character information of user code (new zip code, address display number), The user inputs coordinate information of the new zip code designation area 61 and user code entry candidate area information.

In the keyboard 31, the operator corrects and inputs only the corrected portion in the whole image of the mail P and the user code input area displayed by the display section 32 while referring to the recognition result of the reading section 2. If the recognition result of the reading unit 2 is incorrect, all digits can be corrected and input.

The display section 32 displays the entire image of the mail P inputted by the input interface 26 and the candidate characters of the user code (new post code + address display number) which are the recognition results of the reading section 2. User code input area for inputting the user code, new zip code code specification entry area 6
1 and the like are displayed.

The printer 33 prints out the processing contents of this apparatus and its evaluation table. The voice input unit 34 is for inputting voice recognition target names (town names, names of large recipients, etc.) by voice, and for example, a microphone is used. The voice recognition unit 30 recognizes voice information input from the voice input unit 34.

FIG. 6 shows a memory map of a specific area secured in the first storage unit 24. In FIG. 6, an input mode information area 40 is capable of writing input mode setting information for inputting with the present apparatus. For example, an input user level setting, an image layout change, and the like, which become customizing functions of the present apparatus, can be performed. is there.

In the mail attribute information area 41, the attribute information of the mail P, which is the recognition result of the reading unit 2 for the reject image, is written. As this attribute information, for example,
There are classifications other than postcards / postcards, classifications of handwritten mail / printed mail, and character line direction determination results. In the entire image area 42, a reject image captured by the reading unit 2 is written.

User code description candidate area information area 43
Is the coordinate information of the area where the reading unit 2 recognizes the area where the user code is described with the highest recognition rate except for the new zip code code specification area 61 (that is, the user Code description candidate area information) is written.

New zip code designation area 4
Reference numeral 4 denotes the coordinate information of the new postal code designation area 61 of the postal matter P. New zip code result area 45
Is written as a result of the reading unit 2 recognizing the new zip code from the image in the new zip code designation area 61 of the postal matter P. In the address display number code result area 46, the result of the reading unit 2 recognizing the address display number from the image of the mail P is written.

The image work area 47 is used to promptly perform image processing when the operator looks at the entire image on the display unit 32 and performs image operations such as rotation, reduction / enlargement, and scrolling with the mail P. Area for image processing to be performed.

In the user code input result area 48, the operator visually recognizes the user code by viewing the entire image display area of the display unit 32 or the image display area of the new zip code code designation entry area, and recognizes the user code. Code 3 on keyboard
The result of the key input in step 1 is written.

The search result area 49 is an area in which the user code input by the operator by key input is checked against the address database in the second storage unit 25, and the check result is written. The speech recognition target vocabulary reading information area 50 includes, for example, a town name file 50a as shown in FIG. 7 and a large recipient name file 50b as shown in FIG.
In these files 50a and 50b, recognition target vocabulary reading information (for example, "Akaikai") and corresponding recognition target registration code candidates (for example, 1210XX0) are stored.
3). Further, these are constituted by information of each post office concerning the target post office and information of all the offices collectively including the information.

As a method of registering the post office information, a plurality of recognition target vocabulary reading information can be registered for the same recognition target registration code candidate. Further, different vocabulary reading information such as an abbreviated name or abbreviated name can be registered for the same object. For example, a plurality of "Uratashiyakusho", "Shiyakusho", and "Yakusho" can be registered as vocabulary reading information for "Urata City Office".

In this example, in the case of a street name, as shown in FIG. 7, it is composed of post office unit information (1210XX, 1240XX, etc.) and all station information (1XXXXXX, etc.).
Further, information of another type is also included.

At the time of inputting the code, if the post office is specified, it is searched from the information of the post office unit in the town name file 50a. If the post office is not specified, all the stations in the town name file 50a are searched. Search from information. It should be noted that the same configuration applies to the name of a large recipient.

As shown in FIG. 9, for example, as shown in FIG. 9, the speech recognition target display information area 51 corresponds to a recognition target registration code candidate for each post office corresponding to speech recognition target type information (for a town name, for a large recipient name, etc.). The file name is made up of a file in which the recognition target display information to be recognized is stored, and the file name has a meaning. Hereinafter, an example of the meaning will be described.

That is, the file name is, for example, “high-order 1 digit” and “3-digit / 5-digit code” (for three digits, XX
Is added), and when the upper one digit is "1", it indicates that it is for a street name, and when the upper one digit is "2", it indicates that it is for a large recipient name. In the example of FIG. 9, the file name is for the town name (1210XX) and the station number is "210".
Is shown.

FIG. 10 shows an example of the configuration of the speech recognition section 30. That is, the sound information input from the sound input unit 34 is converted into the characteristic parameter by the sound analysis unit 81. As typical characteristic parameters used for speech recognition, a power spectrum obtained by a band-pass filter or Fourier transform, a cepstrum coefficient obtained by LPC (linear prediction) analysis, and the like are often used. The acoustic analysis unit 81 inputs the feature parameter sequence to the model matching unit 82.

The model collating unit 82 calculates the similarity or distance between each speech model (dictionary information) of the speech to be recognized and the feature parameter sequence, which are created in advance and stored in the speech model storage unit 83. . As a matching method in the model matching unit 82, a speech model is also represented by a feature parameter sequence, and a distance between the feature parameter sequence of the speech model and the feature parameter sequence of the input speech is determined by a DP (dynamic programming) method. Also, a method of expressing a speech model using an HMM (Hidden Markov Model) and calculating a probability of each speech model when a feature parameter sequence of an input speech is input is widely used.

The voice model as the dictionary information stored in the voice model storage unit 83 can be stored in gender units or individual units, and a target voice model is selected from the operator identification information and used. Has become. In this example, for example, a voice model for men and a voice model for women are provided, and the gender is determined based on the input voice information, or the gender information is input at the time of voice input. Is selected.

The recognition result output unit 84 includes a model collation unit 82
The similarity (or distance) with respect to each voice model obtained in (1) is sorted under a certain condition, and the recognition target category having the maximum similarity (or minimum distance) is output as a recognition result.

FIG. 11 shows an example of an operation screen displayed on the display unit 32. An entire image display area 71 for displaying the entire image of the mail P, a new zip code input area 72 and an address display number are shown. It comprises a code input area 73 and is constituted by a user code input area 74 for inputting a user code.

The user code input area 74 includes a new zip code input area 72 and an address display number code input area 73.
Are divided into two different colors, so that the operator can easily recognize them quickly.

The new zip code input area 72 includes an unread digit (reading unit 2) of the recognition result of the reading unit 2 for the new zip code.
The candidate characters to be notified to the operator are displayed based on the pattern in which the characters cannot be recognized).

In the address display number code input area 73, candidate characters as a result of recognition of the address display number by the reading unit 2 are displayed. Next, an operation in the case where a reject mail has occurred in the above-described configuration will be described with reference to flowcharts shown in FIGS.

As an example, a case of a printed type mail item in which the lower four digits of the new zip code are not described in the new zip code code designation description area 61 will be described. For example, FIG.
As shown in (1), the reject mail P handled in this example is a sealed letter whose description format is horizontal writing. The user code of the target mail P is, for example, "210-0001
(New postal code) and 1-2-3-B-201 (address display number), the lower four digits “0001” of the new postal code are not described in the description information, and the reading unit 2 A case in which no recognition result is recognized will be described.

First, in step S1, the entire image of the mail P rejected by the reading unit 2, that is, the mail P whose user code sent from the reading unit 2 via the input interface 26 could not be recognized, Various attribute information of the postal matter P as a recognition result of the reading unit 2, candidate character information of a user code (new zip code, address display number), candidate code description candidate area information, and new zip code designation area The coordinate information 61 is stored in the mail attribute information area 41, the entire image area 42, the user code description candidate area information area 43, the new zip code code designation area information area 44, and the new zip code in the first storage unit 24. The results are written in the result area 45 and the address display number code result area 46, respectively.

Next, in step S2, by reading the entire image of the mail P from the entire image area 42 in the first storage unit 24, for example, as shown in FIG. The whole image is displayed in the whole image display area 71 of the display unit 32. In the case of this example, since no recognition result of the reading unit 2 was recognized, nothing is displayed in the user code input area 74 (the new postal code code input area 72 and the address display number code input area 73). Not done.

Next, in step S3, the operator uses the keyboard 31 to display the entire image of the postal matter P displayed on the display unit 32, and at the normal time, the display unit 32 displays all the user codes. The user code is sequentially input one digit at a time in the 32 user code input areas 74. In the example of FIG.
Since only "10" is described, "210" is first input, and then a town name key for inputting speech recognition target type information (town name) is pressed. When a large recipient name is input as the speech recognition target type information, a large recipient name key is pressed.

Next, in step S4, it is determined whether or not the result input by the operator is a user code or speech recognition target type information. That is, each time the operator inputs the user code by one digit, it is determined whether the user code is the user code or the speech recognition target type information (the name of the town, the name of the large recipient). If the result of this determination is that it is a user code, the flow proceeds to step S5, where the input user code is determined.

If the result of the determination is that the type information is a voice recognition target, the process proceeds to step S6, and the input voice recognition target type information is stored in a first buffer (not shown). At this time, if the current zip code is specified, that is, if the upper three digits of the new zip code are input, the zip code is also stored in the first buffer.

Next, in step S7, it is determined whether the type information of the voice recognition target in the buffer is a town name or a large recipient name. If the name is a town name, the process proceeds to step S8. Proceed to step S9.

After inputting the voice recognition target type information, the operator subsequently inputs a voice recognition target name by voice from the voice input unit 34 (step S14). FIG.
In the case of example 5, the name of the town, "maihama" is uttered.

The voice information thus input is input to the voice recognition unit 30 and subjected to recognition processing (step S1).
5). In this case, as described above, if the operator is male, speech recognition is performed using a male voice model, and if the operator is female, voice recognition is performed using a female voice model.

In step S8, whether or not the current postal code is specified is determined based on whether or not the current postal code exists in the first buffer. If the current postal code is specified, the process proceeds to step S10. move on. In step S10, based on the recognition result of the voice recognition unit 30, in the file 50a for town name (see FIG. 7) of the vocabulary reading information area 50 for voice recognition provided in the first storage unit 24, "1 + current postal code" , And proceeds to step S16.

If the current postal code is not specified in step S8, the process proceeds to step S11. In step S11, based on the recognition result of the voice recognition unit 30, in the file 50a for the town name in the vocabulary reading information area 50 for voice recognition,
Searching from the information area of “1XXXXXX”, step S1
Proceed to 6.

In step S9, similarly to step S8, it is determined whether or not the current postal code is specified, and if specified, the process proceeds to step S12. In step S12, based on the recognition result of the voice recognition unit 30, "2 + current postal code" is stored in the large recipient name file 50b (see FIG. 8) of the vocabulary reading information area 50 for voice recognition.
, And proceeds to step S16.

If the current postal code is not specified in step S9, the process proceeds to step S13. In step S13, based on the recognition result of the voice recognition unit 30, a search is performed from the information area "2XXXXXX" in the large recipient name file 50b of the vocabulary reading information area 50 for voice recognition, and the process proceeds to step S16.

At step S16, steps S10, S
Based on the search processing in step S11, S12 or S13, the first to Nth recognition target registration code candidates are extracted from the speech recognition target vocabulary reading information area 50 and stored in a second buffer (not shown). Proceed to step S17.

In step S17, based on the candidate for the recognition target registration code in the second buffer, the voice recognition target display information area 51 (FIG. 9) provided in the first storage unit 24.
16), the corresponding display information for voice recognition is extracted and displayed above the user code input area 74 on the display unit 32, as shown in FIG. As a display method at this time, the first candidate having the highest recognition accuracy is displayed separately from the other candidates. As a method of displaying the first candidate separately, for example, the first candidate is surrounded by a frame,
It is possible to blink, change the display color, change the display size, change the display position, and the like.

The operator looks at the screen shown in FIG.
An operation of selecting a voice recognition target candidate, an input operation of voice recognition target type information for performing a search again since the target candidate was not obtained in the first search, or an input mode switching operation is performed.

Next, in step S18, the candidate user codes are displayed in the user code input area 74 in real time. That is, the code corresponding to the first voice recognition target candidate is displayed in the user code input area 74 from the beginning, and during the selection of the voice recognition target candidate, the code corresponding to the provisionally selected voice recognition target candidate is displayed. Are sequentially displayed in the user code input area 74.

Next, in step S19, the input result of the operator after displaying the screen of FIG. 16 is monitored, and if a selection code has been input by the operation of selecting a candidate for speech recognition, the process proceeds to step S20. Step S20
Then, the user code corresponding to the voice recognition target candidate selected in step S19 is automatically input, and the user code input in step S5 is determined.

If the speech recognition target type information is input again in step S19, the process returns to step S6, and the above-described operation is repeated. If the input mode switching code is input in step S19,
Proceed to step S21. That is, whether the input of the user code is performed in the key input mode on the keyboard 31 or not.
A mode changeover switch (not shown) for switching between the voice input mode and the input mode is provided. When the voice input mode is stopped and the mode is switched to the key input mode, the mode changeover switch is switched to the key input mode. Thus, the input mode switching code is input.

In step S21, the contents of the first buffer (speech recognition target type information, current postal code, etc.) are referred to, and the flow advances to step S22. In step S22,
A menu screen for key input corresponding to the information in the first buffer referred to in step S21 is displayed on the display unit 32, and the process proceeds to step S23. The key input menu screen has, for example, substantially the same configuration as that of FIG.

The operator looks at the menu screen for key input and selects a speech recognition target candidate, or
Performs a key input operation of a user code. Step S
At 23, the operator's input result after the menu screen for key input is displayed is monitored, and if a selection code is input by the operation of selecting a voice recognition target candidate, the process proceeds to step S20 and the selected voice The user code corresponding to the recognition target candidate is automatically input, and the input user code is determined in step S5.

If a user code has been input in step S23, the flow advances to step S5 to determine the input user code. Hereinafter, the main part will be described using a specific example.

The operator inputs "210" while watching the entire image of the postal matter P displayed as shown in FIG. 15, since only the new postal code "210" is described in this example. Then, press the town name key. Subsequently, in the case of the example of FIG.
And input a voice.

The voice information thus input is recognized by the voice recognition unit 30, and the recognition result is searched for the file 50a for town name (see FIG. 7) in the vocabulary reading information area 50 for voice recognition. In this example, for example, the recognition is performed as “Akaikai → Urasasu → Horie”, and the candidates for the recognition target registration code are extracted as the following voice recognition target candidate information.

Recognition candidate registration code candidate: 1210XX
03,1210XX05,1210XX02 Next, from the extracted candidates for the recognition target registration code,
By searching the voice recognition target display information area 51 (see FIG. 9), the corresponding voice recognition target display information is extracted and displayed on the display unit 32 as shown in FIG. The display of the voice recognition target candidate is displayed above the user code input area 74, and the first voice recognition target candidate “Minghai” is displayed in a frame. This distinguishes it from other speech recognition target candidates.

Further, in the user code input area 74 of the display unit 32, a user code "210-0003" corresponding to the first voice recognition target candidate "Akai" is displayed. Note that as a method of displaying the voice recognition target candidate, index information, a user code, or the like may be displayed next to the voice recognition target candidate. Alternatively, as shown in FIG. 17, an index number may be displayed at the beginning of the candidate for speech recognition, and the index number may be input to select and determine the candidate for speech recognition.

The operator looks at the screen shown in FIG.
In this example, since there is no target voice recognition target candidate, the town name key is pressed again to perform voice input again, and then "maihama" is uttered.

The voice information thus input is recognized again by the voice recognition unit 30, and from the recognition result, the file 50a for the town name in the vocabulary reading information area 50 for voice recognition is obtained.
Is searched. Here, the file for town name 50a is searched, excluding “Akaikai, Urasuyasu, Horie”, which has become a candidate for speech recognition in the previous search. In this example, for example, it recognizes "Maihama → Nekozane → Maehama"
Recognition candidate registration code candidates are extracted as speech recognition target candidate information described below.

Recognition registration code candidate: 1210XX
01,1210XX10,1210XX01 Next, from the extracted candidates for the recognition target registration code,
By searching the voice recognition target display information area 51,
The corresponding voice recognition target display information is extracted and displayed on the display unit 32 as shown in FIG. In this example, since there are a plurality of candidates for the same recognition target registration code, they are displayed as one recognition target registration code.

As shown in FIG. 18, the first voice recognition target candidate “Maihama” is displayed in a frame, and the user code input area 74 is displayed with the first voice recognition target candidate “Maihama”. The corresponding user code "210-0001" is displayed.

The operator selects and determines “Maihama” as a target voice recognition target candidate using, for example, a candidate selection key provided on the keyboard 31 on the screen of FIG. User code "21" corresponding to "Maihama" displayed in
"0-0001" is automatically input and determined as the input user code.

Next, the operator continuously inputs the next address display number (1-2-3-B-201) one digit at a time. The display screen of the recognition target candidates is deleted, and the screen is switched to a user code input screen as shown in FIG.

[0092] The input result of the user code input as described above is written in the user code input result area 48 in the first storage unit 24. Next, the address database stored in the second storage unit 25 is compared with the input result of the user code written in the user code input result area 48 in the first storage unit 24. The collation result is written into the search result area 49 in the first storage unit 24.

If the collation result is correct, the user code in the user code input result area 48 in the first storage unit 24 is transferred to the barcode printing unit 3 and is converted into barcode information. Is printed on the postal matter P to be printed.

If the collation result indicates an incorrect answer, the incorrect answer data is transferred to the barcode printing unit 3 and is also converted into barcode information and printed on the corresponding mail P. As described above, according to the embodiment, the user code is not described, the user code is incomplete,
It is difficult to read the user code, and it is also possible to input a voice recognition target name such as the name of a town or a large recipient for voice mail, such as an incorrect user code. , And display a voice recognition target candidate with high accuracy. The operator can accurately input the user code only by selecting and determining from the displayed voice recognition target candidates.

Also, by providing dictionary information for voice recognition for each gender or individual, the voice recognition rate is improved. Also, the speech recognition target vocabulary reading information area for searching based on the result of speech recognition can have a plurality of recognition target vocabulary reading information for the same target information, and furthermore, the same recognition target registration code Even if a plurality of candidates are extracted, they are displayed as one piece of recognition target display information, so that the accuracy and speed of encoding by voice recognition are improved.

Further, regarding the display of the voice recognition target candidate, the first voice recognition target candidate having the highest recognition accuracy is distinguished from the other voice recognition target candidates so that the operator can quickly and accurately identify the voice recognition target candidate. Selection becomes possible.

The code corresponding to the first candidate for speech recognition is displayed in the user code input area from the beginning, and while the candidate for speech recognition is selected, the temporarily selected candidate for speech recognition is selected. Are displayed in the user code input area one after another, thereby providing an efficient system.

Further, the voice recognition target type information including the presence / absence information of the second or later voice input process can be input to the voice mail to be subjected to the voice recognition. If there is, the word abandonment information extracted as a candidate for speech recognition in the previous speech input is excluded, and it is searched from the vocabulary reading information area for speech recognition and becomes a candidate for a recognition target registration code. A good and accurate system can be realized.

[0099]

As described above in detail, according to the present invention, an operator can input a user code efficiently in a short time and with less fatigue, thereby improving accuracy and labor saving, and speeding up processing. A video coding device that can be realized can be provided.

[Brief description of the drawings]

FIG. 1 is a plan view showing an example of a handwritten mail.

FIG. 2 is a plan view showing an example of a print type mail.

FIG. 3 is a block diagram schematically showing a configuration of a mail processing apparatus to which the video coding apparatus according to the embodiment of the present invention is applied.

FIG. 4 is a block diagram illustrating a configuration of a reading unit.

FIG. 5 is a block diagram schematically showing a configuration of a video coding device according to the embodiment of the present invention.

FIG. 6 is a diagram showing a memory map of a specific area secured in a first storage unit.

FIG. 7 is a view showing a configuration example of a town name file in a vocabulary reading information area for speech recognition in a first storage unit.

FIG. 8 is a diagram showing a configuration example of a large recipient name file in a vocabulary reading information area for speech recognition in a first storage unit.

FIG. 9 is a diagram showing a configuration example of a speech recognition target display information area in a first storage unit.

FIG. 10 is a block diagram illustrating a configuration example of a speech recognition unit.

FIG. 11 illustrates an example of a display screen on a display unit.

FIG. 12 is a flowchart for explaining the operation of the video coding device.

FIG. 13 is a flowchart for explaining the operation of the video coding device.

FIG. 14 is a flowchart for explaining the operation of the video coding device.

FIG. 15 is a diagram showing an example of a specific display screen for a rejected mail on the display unit.

FIG. 16 is a view showing a specific example of a display screen for a rejected mail on the display unit.

FIG. 17 is a diagram showing an example of a specific display screen for a rejected mail on the display unit.

FIG. 18 is a view showing an example of a specific display screen for a rejected mail on the display unit.

FIG. 19 is a view showing an example of a specific display screen for a rejected mail on the display unit.

[Explanation of symbols]

P: mail, 1 ... supply unit, 2 ... reading unit, 3 ... barcode printing unit, 4 ... delay path, 5 ... barcode reading unit, 6 ... sorting unit, 7 ... video Coding equipment,
11: photoelectric conversion unit, 16: character recognition unit, 18: user code recognition unit, 21: CPU, 24: first storage unit, 25: second storage unit, 26: input Interface 27 Keyboard control unit 28 Display control unit 30 Voice recognition unit 31 Keyboard 32
... display unit, 34 ... voice input unit, 50 ... vocabulary reading information area for voice recognition, 50a ... town name file, 50b
... Large recipient name file, 51... Voice recognition target display information area, 61...
62 ... address information area, 71 ... whole image display area, 7
2 ... New zip code input area, 73 ... Address display number code input area, 74 ... User code input area

Claims (12)

[Claims] 1. An overall image on a postal matter in which a user code and address information including at least a postal code and an address display number, and recognition result information on the entire image of the postal matter are fetched. And inputting the user code on the mail based on the displayed image, wherein the entire image of the mail that could not be recognized by the recognition means, and the mail A first input unit for inputting candidate characters of a user code recognized by the recognition unit, an entire image of a mail item input by the first input unit, and a candidate character of the user code; A first display unit for displaying a user code input area for inputting the user code, and a display unit for simultaneously displaying the entire image of the postal matter displayed on the first display unit while viewing the entire image. Second input means for inputting a user code in a user code input area or inputting speech recognition target type information, and voice based on the entire image of the mail displayed on the first display means. Voice input means for inputting a recognition target name by voice; voice recognition means for recognizing voice information input by the voice input means; recognition target vocabulary reading information for each type of voice recognition target; A first storage unit that stores a code, a type of the voice recognition target input by the second input unit, and a search in the first storage unit based on a recognition result of the voice recognition unit. By at least one
Search means for extracting one or more candidates for the recognition target registration code; second storage means for storing recognition target display information corresponding to the candidates for the recognition target registration code; and a recognition target output from the search means By searching the second storage means based on the registration code candidate,
Second display means for displaying at least one or more recognition target display information corresponding to a candidate for a recognition target registration code; and selecting and determining a recognition target candidate from the recognition target display information displayed on the second display means. A video coding apparatus comprising: a recognition target candidate selecting unit. 2. The information stored in the first storage means is:
2. The video coding apparatus according to claim 1, wherein the video coding apparatus comprises information on each post office, and information on all the offices in which the post offices are collected. 3. The video coding apparatus according to claim 1, wherein said first storage means stores a plurality of recognition target vocabulary reading information for the same recognition target registration code. 4. The video coding apparatus according to claim 1, wherein said second display means displays the speech recognition target candidate having the highest recognition accuracy separately from other speech recognition target candidates. 5. The apparatus according to claim 1, wherein the second display unit displays additional information corresponding to the voice recognition target candidate from the recognition target display information near the voice recognition target candidate to be displayed. A video coding device as described. 6. The video coding apparatus according to claim 1, wherein said second display means displays the index information in the vicinity of the voice recognition target candidate to be displayed. 7. The video coding apparatus according to claim 1, wherein said second display means displays, as a single piece of recognition target display information, a plurality of candidates for the same recognition target registration code. 8. The second display means displays an input code corresponding to a voice recognition target candidate having the highest recognition accuracy in the user code input area before selecting a recognition target candidate, and performs recognition. 2. The video coding apparatus according to claim 1, wherein when a target candidate is selected, an input code corresponding to the user code input area is displayed. 9. The recognition target candidate selecting unit selects and determines a recognition target candidate by inputting index information corresponding to the recognition target candidate.
A video coding device as described. 10. The second input means includes an input mode switching means for performing a key input or a voice input and a holding means for holding the voice recognition target type information and post office information. When the input mode is switched by the input mode switching means, the voice recognition means transmits the same information as the voice recognition target type information and post office information held in the holding means of the input mode before the input mode switching. The video coding apparatus according to claim 1, wherein the video coding apparatus is targeted. 11. The search means searches from the information of each post office in the first storage means when the post office is specified, and when the post office is not specified,
3. The video coding apparatus according to claim 2, wherein a search is performed from the all-station information in the first storage unit. 12. The second input means inputs voice recognition target type information including the presence / absence of the second or subsequent voice input processing for the postal matter to be voice-recognized, and the search means comprises: If there is the presence / absence information of the voice input processing,
The recognition target registration code candidate is obtained by excluding the recognition target vocabulary reading information extracted as a voice recognition target candidate in a previous voice input and searching the first storage unit. 2. The video coding device according to 1.