JPS60138688A - Character recognizing method - Google Patents

Abstract

PURPOSE:To avoid a misdecision by obtaining the distance between a feature parameter of the character to be recognized and a dictionary parameter retrieved by a dictionary of different character fonts as well as the mean value of said distance for each dictionary. CONSTITUTION:A feature variable constant Pn is fed and a dictionary DCalpha is retrieved to detect a character which is most approximate to the Pn. The distance between a parameter of the detected character and the Pn is obtained and added to an adder ADalpha, and +1 is given to the contents of a counter CNalpha. Then, a dictionary DCbeta is retrieved to detect a character having the parameter most approximate to the Pn, and this character is added to an adder ABbeta. In the same way, a character having the parameter most approximate to the Pn is detected also with a dictionary DCgamma and then added to an adder ADgamma. Such a procedure is carried out for all characters which are used for discrimination of fonts, and the value Dalpha is obtained by dividing the contents of the adder ADalpha by the contents of the counter CNalpha. While values Dbeta and Dgamma are obtained for adders ADbeta and ADgamma respectively. Then a font corresponding to the minimum one of the values Dalpha, Dbeta and Dgamma is defined as the one which is under recognition.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field] The present invention relates to multi-font character recognition that automatically performs character recognition for a plurality of different types of characters in one font, and particularly to prevention of misrecognition.

[Prior art] For English letters, for example, elite letters and biryoku letters.

There are various typefaces such as italic characters and OCR characters. Therefore, in multi-font character recognition where characters of these various fonts must be recognized, dictionaries corresponding to each font are generally provided. Then, for each recognized character, all character phono 1 to dictionaries are searched, and those with the closest characteristic parameters are selected as candidates, and the closest one among them is made the final candidate.

However, in this type of multi-font character recognition, there are characters that are similar to each other between different fonts.
Some characters may have similar shapes and differ only in size, or may be completely the same. In such a case, the distance between the recognized character and the data in the dictionary will be the same for a plurality of dictionaries with different fonts, and erroneous recognition that may lead to incorrect font selection is likely to occur.

One of the conventional techniques is to count the number of times an unreadable code has been output for each font and use that result to identify the recognition result.However, between fonts with similar parameters, this value is compared. can't identify the font either.

If a font is misrecognized, for example, to reduce search time, the first character l! ! By the result of knowledge.

In character recognition that specifies a dictionary to be used thereafter, erroneous recognition occurs frequently.

[Purpose] The present invention aims to reliably identify fonts and prevent misrecognition.

[Configuration J If a series of character strings to be recognized is composed of only characters of one type of font, for example, the dictionary containing the data with the smallest distance from the feature parameters of the characters to be recognized is Obviously one. In other words, for multiple character patterns sufficient to identify a font, if you search each dictionary to find the closest data, and find the average distance value for each dictionary, you can find the dictionary with the smallest value. can be determined to be a font of a series of character strings.

The average distance value can be obtained, for example, by determining the distance Dn (n=1 to N) for each of N characters, calculating the sum thereof, and dividing the result by N.

If there is a dictionary in which characters within a predetermined distance cannot be retrieved, the value of N may be reduced by -1 for that dictionary without adding Dn.

Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 1 shows a schematic configuration of a character recognition device that implements one embodiment of the present invention. Referring to FIG. 1, character information on a document (not shown) is optically read by a scanner SCA, and photoelectrically converted into binary data for each minute pixel. After this data undergoes a predetermined process for each character pattern in the preprocessing unit PPR, it is applied to the feature extraction unit PPU.The feature extraction unit PPU extracts various feature parameters of the character pattern as numerical data. do.

The search control unit CTR controls each unit, and the feature extraction unit PPU
The dictionary is searched for the character closest to the feature parameter for each character extracted in , and this is used as the recognition result. In this example, three types of character fonts α, β, and T are handled, so the dictionaries D'Cα, DCβ, and D
Cγ is available. The search control unit CTR includes a distance calculation unit CPD that calculates the distance between the feature parameters output by the feature extraction unit PPU and the feature parameters obtained from each dictionary, and three adders ADα, ADβ and ADγ, 3.
Two counters CNα, CNβ and CNγ are connected.

FIG. 2 shows a part of the font discrimination procedure, that is, the operation of the search control unit CTR shown in FIG. This will be explained with reference to FIGS. 1 and 2. First, adders ADα, ADβ and AD
γ, and the contents of counters CNα, CNβ, and CNγ. A counter N for counting the number of characters is set to an initial value, and is incremented by 1 each time a feature parameter is input from the feature extraction unit PPU.

Here, the number of characters used for font identification is limited to Nmax (for example, 10). After inputting the feature parameter Pn, the dictionary DCα is first searched to find the character closest to Pn. The distance between the parameter of the searched character and Pn is determined and added to the adder ADα, and the content of the counter CNα is incremented by 1. If a character whose distance from the feature parameter Pn is less than a predetermined value cannot be searched (if it does not exist in the dictionary), the contents of the adder ADα and the counter CNα are left as they are.

Next, the dictionary DCβ is searched to find the character whose parameter is closest to Pn. The distance between the parameter of the searched character and Pn is determined, and the result is added to the adder ADβ, and the content of the counter CNβ is incremented by 1. If a character whose distance from the feature parameter Pn is less than a predetermined value cannot be retrieved, the contents of the adder ADβ and the counter CNβ are left as they are. Similarly, the dictionary DCγ is searched for the character whose parameter is closest to Pn, and that parameter and P
The distance to n is determined, and the result is added to adder ADγ, and the content of counter CNγ is incremented by 1. If a character whose distance from the feature parameter Pn is less than a predetermined value cannot be retrieved, the contents of the adder ADγ and the counter CNγ are left as they are.

If the above operation is continued until N>Nmax, the adders ADα, ADβ, and ADγ each have the font α
, β, and T dictionaries and each of Nmax recognized characters is stored, and counters CNα, CNβ, and CNγ actually input distance data to adders ADα, ADβ, and ADγ, respectively. The number of times it has been performed is stored.

Therefore, when Nmax readings are completed, the adder A
A value Dα obtained by dividing the contents of Dα by the contents of the counter CNα, and a value D obtained by dividing the contents of the adder ADβ by the contents of the counter CNβ.
A value Dγ is calculated by dividing the contents of β and the addition @AD7 by the contents of the counter CNγ.

Each value Dα, Dβ and Dγ is the font α
, β and γ, and the parameters of Nmax recognized characters.

Therefore, there is a large difference between Dα, Dβ and Dγ. That is, for example, if the fonts of the character strings being recognized are all α, the data in the dictionary DCα is Nrnax
Since the distances are small for all the characters, the average value Dα is also very small, but for the dictionaries of other β and γ fonts, the distances are small for some characters because the parameters are similar. However, since the distance is large for other characters, D
β and Dγ have considerably large values compared to Dγ.

That is, by comparing the values of Dα, Dβ, and Dγ, the font corresponding to the smallest value is the font of the character being recognized. Therefore, once this determination is completed, only the dictionary containing the determined font is selected, and searches in other dictionaries are prohibited. In this way, after the font determination, only one dictionary needs to be referred to, so that the search process can be performed in a short time.

In the above embodiment, only the first Nmax characters are subject to font determination, but depending on the type of document, characters of multiple fonts may appear in one document.

In such a case, the above method may cause erroneous recognition during the process.

Next, another embodiment of the present invention will be described. FIG. 3 shows a schematic configuration of a character recognition device implementing the present invention in another aspect, and FIG. 4 shows its font determination procedure. Third
Referring to the figure, in this embodiment, the search control unit CTR includes three sets of memories Mα.

Mβ and M7 are connected. Other configurations are first
Same as figure.

This will be explained with reference to FIG. As in the previous embodiment, first read N max characters, search the dictionaries of fonts α, β, and γ, calculate the distance of the nearest character, input it to each adder, and add 1 to each counter. do. In addition, in this embodiment, when the search in each dictionary is completed, the searched character (in each dictionary) with the closest distance,
That is, the code data of the first candidate character is stored in the address specified by N of the memory Mα2Mβ or Mγ associated with each font.

Therefore, Nmax first candidate characters of each font are stored in the memories Mα2Mβ and Mγ. N
When the max first candidate character searches are completed, the average distances Dα and Dβ of each font are calculated in the same manner as in the above embodiment.
and Dγ, and find the smallest one by comparing them. In this example, the memory (Mα9Mβ or M
The content of γ) is read out, and each of the Nmax first candidate characters therein is set as a final candidate character and output as a recognition result.

When this is completed, the counter N is set to the initial value again and the above operation is repeated. That is, in this example, Nma
The character font is determined for each x number of characters, and Nmax first candidate characters stored in memory in advance are retrieved.

In the above embodiment, the characters used for font discrimination are specified as the numerical value Nmax, but for example, 1
It may be specified in character strings for lines or in units of one page.

In addition, in the font determination procedure shown in Figure 2, font determination is not performed after first performing font determination for Nmax characters, but 1. For example, when a page of a manuscript is updated, N is returned to the initial value and font determination is performed. You may change it to

Note that in the embodiments shown in FIGS. 1 and 2, the adder, counter, memory, etc. are each made into independent components, but when using a microcomputer, for example, it is possible to use the functions of the adder, counter, memory, etc. , the present invention can be implemented without adding any special components.

[Effects] As described above, according to the present invention, for a plurality of characters,
Since the font is determined based on the average distance between the data in each dictionary and the characteristic parameter of the character, misjudgment is eliminated.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of a character recognition device implementing one aspect of the present invention. FIG. 2 is a flowchart showing a font discrimination procedure in the apparatus shown in FIG. FIG. 3 is a block diagram showing the configuration of a character recognition device implementing another embodiment of the present invention. FIG. 4 is a flowchart showing a font discrimination procedure in the apparatus of FIG. DCα, DCβ, DCγ: Dictionaries ADα, ADβ, ADγ: Adders CNα, CNβ, CNγ: Counters Mα2Mβ, MT: Memory

Claims (3)

[Claims] (1) In character recognition that targets multiple types of character fonts; For multiple characters, calculate the distance between the characteristic parameters of the recognized characters and the dictionary parameters searched in dictionaries of different character fonts, and calculate the distance. A character recognition method characterized in that an average value is determined for each dictionary, and the results are compared with each other, and a font of the dictionary is specified according to the result to obtain a recognition result. (2) For the first part of characters, calculate the distance between the feature parameters of the recognized characters and the dictionary parameters searched in different character phono 1- dictionaries, and calculate the average value of the distance for each dictionary. The character recognition method according to claim 1, wherein a font of a dictionary to be searched thereafter is specified according to the result of comparing the results with each other. (3) For multiple characters, calculate the distance between the characteristic parameters of the recognized character and the dictionary parameters searched in dictionaries of different character fonts, store the searched characters in each dictionary of each font, and calculate the distance. The above patent claim specifies the font 1- according to the result of determining the average value of for each dictionary and comparing the results with each other, and uses the search character obtained in the dictionary of the specified font as the recognition result. The character recognition method described in item (1).