JPS6024771A - Address recognition system - Google Patents

Abstract

PURPOSE:To improve the recognition accuracy by decreasing the word dictionary size when the word is recognized, by recognizing the address of the mail, extracting the word from the classified address name by stages. CONSTITUTION:When the mail of the picture signal transferred from the facsimile receiving machine, etc., is given to a input terminal 10, a picture signal receiving section 11 receives the mail of the picture signal by means of the command of a control section 12 and stores thin in a mail memory section 13. After the storing ends, a character pattern recognition section 14 extracts the address picture signal out of the mail of the picture signal stored in the mail memory section 13 and recognizes characters at read field unit. Next, an address format definition section 15 controls a field selection section 16, a word extraction section 17 and an address recognition section 18 based upon the address format definition table stipulating the class composition of the address name memorized beforehand, recognizes the address name and sends the mail through a mail transmission section 20 to the desired transmission destination.

Description

[Detailed Description of the Invention] (Field to which the invention pertains) The present invention recognizes the characters of a hierarchical destination name included in an image signal mail, and uses a pre-stored hierarchical word dictionary to gradually This invention relates to a destination recognition method for controlling mail destinations by performing word recognition.

(Prior Art) Conventionally, in order to send mail with image signals for different destinations to a desired destination using destination recognition with a character recognition function,
2. Description of the Related Art A transmission processing method is provided that allows a large amount of mail to be input to different destinations at once and efficiently transmitted to the desired destination.

Figure 1 is an explanatory diagram showing an example of application of this transmission processing method, in which a sheet of paper with mail A and its addressee σ handwritten on it, and a sheet with handwritten mail B and its addressee written on it are sent to a facsimile transmitter. When faxing is done manually, the processing device C to which this method is applied recognizes the characters on the paper on which address A is written and sends it to address a.

The destination of b is recognized and the desired facsimile S is sent based on this.
This shows that mail A and mail B corresponding to IJ receivers PAX a and FAX b are sent to IB.

In the second application example, when the number of destinations increases, the size of the word dictionary that stores the destination names increases, which has the disadvantage that the recognition accuracy of destination recognition decreases rapidly. Due to the increase in the number of characters, the number of characters required to write one destination name increases, which has the disadvantage that recognition accuracy often decreases due to writing errors.

(Object of the Invention) In order to solve these drawbacks, the present invention is characterized in that it separates mail destinations by extracting and recognizing words step by step from hierarchical destination names. The aim is to reduce the size of the word dictionary H1l during recognition, thereby improving the recognition level and degree of destination recognition.

(Structure and operation of the invention) FIG. 2 is an explanatory diagram showing an example of a destination writing sheet used in the present invention.

1 is a destination entry form, and there are two ways to write the destination name in a hierarchical format. The figure (ct) is an example in which there is only one reading field 2, which is the output unit of the character recognition result, and a hierarchical destination name is written in the reading field 2. FIG. 2A shows an example in which reading fields 2-1.2-2.2-3 are provided and filled in for each hierarchy.

FIG. 3 shows the configuration of the reading field.

In the same figure <a), in the only reading field 2, words 3-1. #7'7 indicates that it will be completed. <b> in the same figure shows a case where reading fields 2-1.2-2.2-3 are provided for each layer, and words 3-1.3-2 and 3- of the first, second and third layers are provided. 3rd
They correspond to reading fields 2-1, 2-2 and 2-3, respectively, and the end of each word is composed of a delimiter character 4 and a blank character 5 following this.

FIG. 4 is a block diagram showing the configuration of an embodiment of a destination recognition control device to which the method of the present invention is applied, in which 1o is an input terminal;
12 is a control unit, 13 is a mail memory unit, 14 is a character pattern recognition unit, 15 is a destination person definition unit, 16 is a field selection unit, 17 is a word extraction unit,
18 is a word recognition section, 19 is a destination control section, 2o is a mail transmission section, and 21 is an output terminal.

Next, its operation will be explained. When an e-mail containing a g (compressed image signal) transferred from a facsimile transmitter or the like (not shown) is given to the input terminal 1o, the image signal receiving section 11 receives the image signal by a command from the control section 2. The image signal mail is received and stored in the mail memory 13. After the storage is completed, the character pattern recognition section 14 selects the second image signal mail from among the image signal mail stored in the mail memory 13 according to a command from the control section 12. The image signal of the address writing sheet 1 explained in the figure is extracted and character recognition is performed for each reading field.

Next, the destination format rectangle section 15 receives the command from the control section ]2, and based on the destination format definition table that defines the hierarchical structure of destination names stored in advance, the field selection section]6 and the word extraction section 17 and controls the word recognition unit 18. In accordance with the instructions from the destination date mat definition section 15, the field selection section 16 selects a character/kucoon recognition section]
A desired reading field 2 or 2-1.2-2.2-3 is selected from among the reading fields output in step 4. After the selection is completed, the word extraction section 17 converts the destination formant definition section 1 to the destination formant definition section 1.
According to the instructions in step 5-C, first extract the word delimiter 4 for the selected reading field 2 or 2-1 to 2-3, and then extract the word delimiter 4 based on the extraction result of the delimiter 4. Extract words. When word extraction becomes redundant, the word recognition section 18 recognizes the words outputted from the word extraction section 17 word by word according to instructions from the destination format definition section 15. After word recognition is completed, destination nail definition section 1
According to the instructions in step 5, if the words in the lower hierarchy are included in the same reading field, the word extraction unit 17 operates again, and if the words in the lower hierarchy are included in a different reading field, the field selection unit 16 operates again. This series of operations is repeated until the lowest level word is recognized.

Destination format definition? : When the series of controls by u15 is completed, the destination control section 19 receives a command from the control section 12 and converts the mail by referring to the destination name-to-destination conversion table stored in advance based on the destination name recognized in the word recognition section 8. Control destinations.

The destination information uses a moving bed ID number, telephone number, etc. When the control operation of the destination control unit 19 becomes possible, the control unit 120
In response to the command, the mail transmitting section 20 cuts out the image signal for each mail, and sends it to the output terminal 2 according to the control instruction of the destination control section 190.
1 to the desired destination.

In addition, although the example of the case where the number of layers is three is shown in "Second Notes Explanation", the number of layers is not limited to this, and even higher-order one
It is clear that it can be applied to the bamboo layer.
However, since the addresses used in on-campus large-scale communication are difficult to use, and there are many forms of names, it is very effective to apply this method to this field.

(Effects) As explained above, the present invention controls mail destinations by extracting and recognizing words from hierarchical destination names step by step. Not only does it have the advantage of avoiding a decline in recognition accuracy, but it also saves labor in reject processing operations that would be required due to a decline in recognition accuracy. It has the advantage that it can be sent to a desired destination.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram showing an example of application of a conventional transmission method, FIG. 2 is an explanatory diagram showing an example of address writing paper used in the present invention,
FIG. 3 is a diagram showing the configuration of a reading field, and FIG. 4 is a block diagram showing the configuration of an embodiment of the IL/ζ destination recognition control device to which the method of the present invention is applied. 1 ・・・・・・Destination entry form, 2.2-1.
2-2.2-3...reading, 3-
1-3-3... Words from the 1st to 3rd hierarchy,
4 ・・・・・・ Delimiter character, 5 ・・・・・・
...Blank character, 10...Input terminal, 11
...... Image signal receiving section, 12...
Control unit, 13... Mail memory unit, 14.
...Character pattern recognition section, ]5...
...Destination format definition part, J6...Field selection part, 17...Word extraction part,
18・・・・・・・Word recognition section, 19・・・・・・・
・Destination control unit, 20...Mail transmission unit, 2
1... Output terminal. Patent applicant Nippon Telegraph and Telephone Public Corporation Figure 1 Figure 2 Figure 3 ((11

Claims (1)

[Claims] An image signal receiving section that receives an input image signal mail;
a mail memory section that stores the image signal; a character recognition section that extracts and recognizes a predetermined character pattern from the mail stored in the mail memory section; In a document communication processing device comprising a mail sending unit that sends a mail image signal to a desired destination, and a control unit that generates a timing signal necessary for each circuit, the above-mentioned character pattern recognition unit A field selection section that selects a desired reading field from the reading fields that are output units of recognition results, and a word extraction section that divides the reading field selected by the field selection section into r word units and extracts desired words. a word recognition unit that recognizes the extracted word; and a destination format definition that controls the field selection unit, word extraction unit, and word recognition unit based on a destination format definition table that defines a 1+¥ layer structure of a destination name. and a destination control section that controls the destination of the mail of both signals from the word identified by the word recognition section based on the destination format definition table, l! A destination recognition method that uses verbal cues to control the destination by extracting and recognizing words step by step from the layered destination name.