JP2588261B2 - Address database search device by OCR - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial application field) The present invention relates to a data search for searching data corresponding to a desired keyword input from a plurality of data each having a plurality of keywords. Related to the device.

In particular, the present invention relates to an address database search device by OCR that searches a database using European address information as a keyword.

(Prior Art) This type of data search device has a table storing data corresponding to each keyword, and searches for a keyword that matches the input keyword from all the keywords stored in the table. Then, data corresponding to the searched keyword is output. There is not only one keyword for identifying each data, but usually a plurality of keywords for each data. For example, in the address of the English letter, the keyword for the state of the United States, New York, was changed to multiple names such as the official name "NY", the full spelling "New York", or the abbreviation "NY" Since there are keywords, it is necessary to store data as New York for each of the plurality of keywords.

Since a plurality of keywords are present for each data as described above, the operation of searching for a keyword that matches the input keyword needs to be performed for all the keywords that exist for each data. Therefore, the search time is several times the number of keywords that exist in comparison with the total number of data.

(Problems to be Solved by the Invention) As described above, when a plurality of keywords exist for each data, the search operation has to be performed for all the keywords in the related art, so that the search time is greatly increased. There is a problem.

The present invention has been made in view of the above, and an object of the present invention is to provide an OCR-based address database search device capable of performing a search in a short time and efficiently.

[Structure of the Invention] (Means for Solving the Problems) In order to achieve the above object, an address database search device based on OCR of the present invention uses address information based on a character recognition result based on OCR of a Western address of a postal item as a keyword. A keyword that represents a state name is a two-letter official name "standard state name search table", three or more characters of a "full spell state name search table", and a combination of one character Keyword classification storage means for classifying and storing in the abbreviation "abbreviated state name search table", and the number of characters from the character string of the state name input as a result of the character recognition by the OCR. ,
State name search table determining means for determining whether the number is two characters, three or more characters, or a combination of one character, and the state name search table stored in the one state name search table determined by the determining means. And a search unit that searches for a keyword that matches the input keyword from among the keywords that have been input, and that outputs address data corresponding to the searched keyword.

(Operation) In the data search device of the present invention, a plurality of keywords are classified and stored for each keyword classification information, and the keyword classification information corresponding to the input keyword is determined, and the keyword classification information corresponding to the determined keyword classification information is determined. The matching keyword is searched from the stored keywords.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 shows a keyword information table 1 and a search table 3a used in a data search device according to an embodiment of the present invention.
FIGS. 3A to 3N are explanatory diagrams showing a data table 5; The keyword information table 1 is a table that stores keyword classification information for determining which of the keyword classification information corresponds to the input keyword. This keyword classification information is used for all of the classified keywords. This is common information. For example, the keyword separation information stored in the keyword information table 1 is information characterizing each keyword. This is information such as a common character string in the keyword, for example, a common character string in which the first character of the keyword starts with “A”. Usually, a database often uses keywords having similar characteristics to some extent, and such information can be easily extracted when a search table is created.

Search tables 3a to 3n are provided respectively corresponding to the keyword classification information stored in the keyword information table 1. That is, the search table 3a stores each keyword classified by the keyword classification information stored first in the keyword information table 1,
A pointer indicating the address of the data corresponding to each keyword is shown on the right side corresponding to each keyword.
In addition, the search table 3b stores each keyword classified by the keyword classification information stored second in the keyword information table 1 and the corresponding pointer. Similarly, the search table 3n stores the keyword in the keyword information table 1. Each keyword classified by the keyword classification information stored at the n-th position and a corresponding pointer are stored.

The pointers stored in each of the search tables 3a to 3n indicate the address of each data stored in the data table 5, and by reading out the data at this address, it is possible to obtain data for the searched keyword. You can.

FIG. 2 is a flowchart showing a search operation using each table shown in FIG. In the figure, first, a keyword information table 1
Extract the first keyword classification information (step 11)
0), the number of the search table is set to 1 corresponding to the first keyword classification information (step 120), and it is checked whether the keyword classification information matches (step 130). If they do not match, the operation of extracting the classification information of the next keyword from the keyword information table 1 and incrementing the number of the search table (steps 140 and 150) and checking whether or not the keyword classification information matches is repeatedly performed (steps). 160).

If there is matching keyword classification information, a search table corresponding to the matched keyword classification information is selected (step 180), and the data number indicated by the pointer of the first keyword in the search table is set to 1 (step 180). 190), it is checked whether the first keyword in the search table matches the input keyword (step 200).
If the keywords do not match, the operation of incrementing the data number (steps 210 and 220) and checking whether the next keyword in the search table matches the input keyword is repeated (step 230). If the keyword matches, the data of the pointer corresponding to the keyword, that is, the data of the data number shown in step 190 or 220 is output as the data for the input keyword.

As described above, a plurality of search tables are provided separately according to the keyword classification information, and only the search table of the keyword classification information corresponding to the input keyword is searched, whereby the search time by the keyword can be significantly reduced. Specifically, the total number of keywords is N
Assuming that the number of keywords in the selected search table is M, the search time is reduced to N / M.

FIG. 3 is a block diagram showing the configuration of another embodiment of the present invention. FIG. 1 shows an example of a database search using OCR, in particular, a system that reads data described on a letter, searches an address database, and sorts the letter. In the figure, a letter 21 conveyed along a letter conveying path 11 is read by an optical system 13, A / D converted, and supplied to a character recognition unit 15 as an image pattern signal. The character recognizing unit 15 recognizes the character by, for example, the compound similarity method or the like, and supplies the character to the database search unit 17 as, for example, an ASCII code together with other address information, for example, the position information of the character. The database search unit 17 generates word data of the data of the letter 21 using the supplied character code and character position information, and searches the address database 23 using the word data as a keyword. Then, the address written in the letter 21 is read by searching the address database 23, and an address code corresponding to, for example, a postal code is generated. This address code is supplied to the transport control unit 19. The transport control unit 19 supplies a stacker control signal corresponding to the address code to the stacker unit 25, and stores the letter 21 in a predetermined stacker.

As shown in FIG. 4, the letter 21 shows a letter in the United States as an example. As shown in the figure, the address 31, the house number 33 of the destination address, the street 35, the city name 37, the state name 39, and the ZI
Although P41 is shown, the database search unit 17
41, state name 39, city name 37, street name 35, house number 33
And searches the address database 23 in this order to generate a corresponding address code.

In the letter 21, for example, the official name "NY" is described as the state name 39, which is "New Yor
There are several variations, such as "k" or spaced apart from the abbreviation "NY". Therefore, a search for the state name would return "New York State"
The following keywords are required.

NY New-York N_Y (Caution _ indicates a space) Here, the keyword of the standard name shown is two characters for all states, and is an abbreviation of one character combination.
It can be seen that the full spelling consists of a combination of words of three or more characters. Therefore, as described above, when "NY" is input as a keyword, only the standard type state name table based on the number of characters needs to be searched.

FIG. 5 shows a keyword information table 5 having such a structure.
0, search tables 51, 52, 53 and a data table 55 of a city name list pointer are shown. The keyword information table 50 has two characters, three or more characters, and one character as the keyword classification information. For each of these keyword classification information, a standard state name search table 51, a full spell state name search table 52, a state abbreviation state name A search table 53 is provided.

Therefore, as described above, when the two letters “NY” are described in the letter 21, when the letters “NY” are read,
Since it is known that the characters are two characters, the keyword information table 50 is subtracted, and the standard state name search table 51 is selected. Then, this search table is searched with a keyword to obtain a pointer to the city table. With this configuration, it is not necessary to search a full-spelled or abbreviated table, so that high-speed search can be performed.

[Effects of the Invention] As described above, according to the present invention, a plurality of keywords are classified and stored for each piece of keyword classification information, and keyword classification information corresponding to the input keyword is determined. Since the matching keyword is searched from the keywords stored corresponding to the determined keyword classification information, it is not necessary to search all the keywords, and the keyword is stored corresponding to the determined keyword classification information. Since you only need to search for keywords,
The search time can be significantly reduced, and the speed and efficiency can be improved.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing the configuration of a search table used in a data search device according to an embodiment of the present invention, FIG. 2 is a flowchart showing a search operation using the search table of FIG. 1, and FIG. FIG. 4 is a block diagram showing a configuration of another embodiment of the invention, FIG. 4 is a diagram showing a letter used in the embodiment of FIG. 3, and FIG. 5 is a configuration of a retrieval table used in the embodiment of FIG. FIG. 1 Keyword information tables 3a to 3n Search table 5 Data table 15 Character recognition unit 17 Database search unit 21 Letters 23 Address database

Claims (1)

(57) [Claims] An address database search device for searching a database using address information as a keyword based on a character recognition result by OCR of a Western address of a postal matter, wherein a keyword representing a state name is expressed by a two-letter formal name. Keyword classification storage that is classified and stored in a certain "standard state name search table", a "full spell state name search table" of three or more characters, and an "abbreviated state name search table" which is an abbreviation of a combination of one character. Means for determining whether the number of characters is two, three or more, or a combination of one character from the character string of the keyword representing the name of the state input as a result of the character recognition by the OCR. State name search table judging means for judging, and the input from the keywords stored in one state name search table judged by the state name search table judging means The search keywords that match the keywords, OCR address database search device according to, characterized in that it comprises a search means for outputting the address data corresponding to the search keyword.