JPH01296393A - Category deciding device - Google Patents

Abstract

PURPOSE:To discriminate the category of words having variable lengths by calculating evaluated values of characters at every kind of character and the evaluated value of a word from the sum of the number of each digit of the calculated value and deciding the category composed of the numbers of kinds and digits of the characters constituting the word. CONSTITUTION:The address information written on a postal matter P is read by a scanner 11 and sent to a recognizing section 13 after the character of the information are segmented at a character detecting and segmenting section 12. At the recognizing section 13, capital alphabetical characters, small alphabetical characters, numerical characters, and special characters are recognized from the inputted characters based on character information stored in the 1st memory 14 and a recognized result is sent to a category deciding section 15 together with character patterns constituting words. The deciding section 15 discriminates whether each word is constituted of numerical characters or alphabetical characters only or a combination of alphabetical and numerical characters and how many digit each word has. At a word recognizing section 17, the inputted address information is separated into 'alphabetical' and 'numerical' words and whether a corresponding word represents a 'state', 'city', etc., is recognized based on words stored in a 3rd memory 18.

Description

[Detailed Description of the Invention] [Purpose of the Invention (Industrial Application Field) This invention is applied to a so-called address reader that reads address information attached to foreign mail, for example, and is applicable to a variable-length address reader. The present invention relates to a category determination device that can perform category determination even on words.

(Prior Art) For example, a postal code assigned to a domestic mail item is composed of a combination of character types including handwritten numbers, printed numbers, a 〒 mark, and a - (hyphen). This postal code combination category determination is based on the combination of character types that has the maximum similarity value of the composite similarity calculated for each character type that makes up the postal code: 〒 mark, 3 digits, and 3 digits. This is done by determining which format matches - (hyphen), 2-digit numbers, or only 3-digit numbers.

In other words, conventional category determination is based on the 1 that constitutes address information.
The combination of character types with the maximum similarity in two words is
It was impossible to make a judgment unless it matched the category to be judged. Therefore, it has been impossible to perform category determination on variable length words made up of multiple character types, such as address information attached to foreign mail.

Further, in conventional devices, if there is even one blurred character in a word, for example, it is impossible to perform category determination.

(Problems to be Solved by the Invention) This invention solves the problem that it is not possible to perform category determination for words of variable length consisting of multiple character types. The present invention aims to provide a category determination device that can reliably determine the category of words of variable length.

[Structure of the Invention] (Means for Solving the Problems) This invention comprises a reading means for reading words of variable length consisting of one or more digits in at least two types of characters, and a word is constructed from the reading output of the reading means. a recognition means for recognizing characters; a first calculation means for calculating an evaluation value for each digit of a character type constituting a word from a recognition output of the recognition means; A second calculation means that calculates a word evaluation value made up of these character types by adding each digit, and a category made of the character types and number of digits that make up the word is determined from the calculated word evaluation value. and determining means.

(Operation) The present invention calculates an evaluation value for each character type constituting a word by the first calculation means from the characters read by the reading means and recognized by the second reading means, and calculates the evaluation value for each character type constituting the word by the second calculation means. A word evaluation value is calculated by calculating the sum of the evaluation values calculated for each digit, and from this calculated word evaluation value, the determination means determines the category consisting of the character type and the number of digits of the character that make up the word. By doing so, it is possible to determine the category of variable-length words made up of multiple character types.

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

In FIG. 1, address information written on a postal item P is read by a scanner device 11, and this read address information is supplied to a character detection and cutting section 12. The character detection and cutting section 12 cuts out characters constituting the address information, and the cut out character information is supplied to the recognition section 13. In this recognition unit 13, based on the character information stored in the first memory 14, the input character is "
It recognizes whether it is an alphabetic character, a lowercase alphabetic character, a number, or a special symbol.

By the way, the address information attached to foreign mail is
As shown in the figure, it is composed of "alphabet letters", "numbers", and "special symbols" such as periods and hyphens.

Furthermore, these words consisting of "alphabetic letters,""numbers," and "special symbols" do not have multiple character types mixed together in one word. For example, words consisting only of "alphabetic letters" such as rNEWJ, A word consisting only of ``numbers'', such as ``r7THJ'', or a word consisting only of ``digits'' from the beginning, such as ``r7THJ'' (
In this case, it is composed of words arranged in such a way that up to one digit (1 digit) is a ``number'' and after that is an ``alphabetic letter.'' For example, the words ``number,'' ``alphabetic letter,'' and ``number'' do not appear in one word. It is.

The recognition unit 13 recognizes "alphabetic characters", "lowercase letters", "
Recognition results (candidates) for ``numbers'' and ``special symbols'' are output for each word, and the recognition results are supplied to the category determination unit 15 together with the character patterns that make up the word.

This category determining section 15 is constituted by, for example, a microcomputer, and this category determining section 15 includes:
A second memory 16 in which operating programs and the like are stored is connected.

This category determination unit 16 determines whether the read word is composed only of "alphabetic characters" or "numbers".
By combining numbers and letters, it is determined whether these character types are composed of the first digits.

FIG. 3 shows the operation of the category determining section 15, and the operation of the category determining section 15 will be described below with reference to FIG.

For example, if the recognition result for a 6-digit word is input from the recognition unit 13, a list as shown in FIG. 4 is created according to the recognition result of the characters that make up this word (step 5TI). .

In other words, Fig. 4(a) shows the recognition result of recognizing a 6-digit word as a "number", where the part that was recognized as a "number" is marked with an "O", and the part that cannot be recognized is marked with an "O". rX
J is set corresponding to each digit.

In addition, in FIG. 6B, similarly, for a six-digit word, "O" is set for the part that can be recognized as an "alphabet character", and "x" is set for the part that cannot be recognized.

Next, based on the list created in this way, it is determined whether the category is "alphabetic characters" only, "numbers" only, or a combination of "alphabetic characters" and "numbers".

First, regarding the list of "numbers" shown in Figure 4(a),
A list is generated by counting the number of rOJs for each digit, as shown in FIG. 5(a), with rOJs as 1 point and "x" as 0 points from the beginning of the word (step 5T2).

On the other hand, for "alphabetic characters", from the end of the list shown in Figure 4(b), rOJ is set as 1 point and "×" is set as 0 points, and the list obtained by counting the number of rOJ for each digit is shown in Figure 5(b). It is generated as shown in (step 5T3).

From the evaluation values obtained for each digit in this way, as shown in Figure 5, "6 digits of alphabets" and "0 digits of numbers", "5 digits of alphabets" and "1 digit of numbers"... "1 digit of alphabets" ” and “5-digit number”,
By adding the evaluation values of ``numbers'' and ``alphabetic characters'' for the combinations of ``0-digit alphabetical character'' and ``6-digit number,'' the word evaluation value as shown in Figure 6 is obtained (Step 5
T4). It is determined whether or not there is a value whose maximum value is larger than the slice value among the obtained word evaluation values (step 5T5), and as a result, a "number" having the maximum value exists.
The combination of "alphabetic characters" is output to the word recognition unit 17 as the desired category (step 5T6).

In the case shown in FIG. 6, the maximum value of the word evaluation value is "5", which is larger than the slice value "4", so this word evaluation value "
``4 digits + 2 digits alphabet'' corresponding to ``5'', ``5 digits + 5 digits''
It can be determined that "1 digit alphabetic character" is the category of the input word, and the category thus determined is supplied to the word recognition unit 17.

In this word recognition unit 17, according to the determined category, the input character information is a word consisting of "alphabetic characters",
It is divided into words consisting of "numbers," and based on the words stored in the third memory 18, it is determined whether the corresponding word is "state name", "city name", etc. What is there is recognized. This recognition result is supplied to a postal code determination unit (not shown), and the recognized "state name" and "
The corresponding "zip code" is determined from the city name, etc.

According to the above embodiment, the evaluation values of "alphabets" and "numbers" included in the words constituting the address information read from the mail are obtained, and the evaluation values from the beginning of "alphabetic characters" and the evaluation values of "numbers" are calculated. A word evaluation value is obtained by calculating the sum of each digit of the evaluation value from the end, and if this word evaluation value is greater than a predetermined value, the categories of "numbers" and "alphabets" corresponding to that word evaluation value are determined. , the word is output as a category. Therefore, even if the word is a variable length word made up of multiple character types, the category of the word can be reliably determined.

Furthermore, since character evaluation values and word evaluation values are obtained, it is possible to improve the accuracy of category discrimination even when a read word contains blurred characters.

Furthermore, since word evaluation values are being calculated, if multiple word evaluation values among these word evaluation values are larger than the slice value, the maximum value of these evaluation values is determined to be the one that corresponds to the word category. Second, even if a plurality of word evaluation values exceed the slice value at the same time, categories can be determined more reliably.

Note that this invention is not limited to the above embodiments,
Of course, various modifications can be made without departing from the gist of the invention.

[Effects of the Invention] As detailed above, according to the present invention, from the characters read by the reading means and recognized by the recognition means,
Calculating the word evaluation value by calculating the evaluation value for each character type constituting the word by the first calculation means, and calculating the sum of the evaluation values calculated for each character type by the second calculation means for each digit, From this calculated word evaluation value, by determining the category consisting of the character type and the number of digits of characters constituting the word using the determination means, it is possible to determine the category of a variable length word composed of multiple character lines. A possible category determination device can be provided.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a diagram showing an example of characters to be read, and FIGS. 3 to 6 are diagrams for explaining the invention in detail. . 11... Scanner device, 12... Character detection cutting section,
13... Recognition unit, 15... Category determination unit. Applicant's agent Patent attorney Takehiko Suzue 112 Figure 3

Claims (1)

[Scope of Claims] Reading means for reading words of variable length consisting of one or more digits in at least two types of characters; recognition means for recognizing characters constituting a word from the read output of this reading means; A first calculation means that calculates an evaluation value for each digit of the character types that make up a word from the recognition output, and a A second calculating means for calculating a word evaluation value, and a determining means for determining a category consisting of a character type and a number of digits forming a word from the calculated word evaluation value. category determination device.