JP3086264B2 - Character space recognition method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition apparatus for processing Japanese sentences, and more particularly to a character recognition apparatus of this kind which distinguishes a space between characters (space character) into two-byte / one-byte characters. The present invention relates to an inter-character space recognition method.

[0002]

2. Description of the Related Art Conventionally, a method of recognizing an inter-character space (space character) in an English sentence OCR or the like is to calculate a character print pitch and compare the print pitch with the width of a space between characters. It has been common practice to determine the presence or absence of a space in a blank portion and determine the number of spaces.

[0003]

In the case of Japanese sentences, half-width numbers and alphabets are often used in addition to full-width kanji and hiragana, and the space is also a mixture of full-width and half-width spaces. Often used. In recognizing a sentence in which full-width and half-width spaces are mixed, it is necessary to recognize and output a full-width space and a half-width space separately.

However, there are various inconveniences when a conventional inter-character space recognition method is applied to Japanese sentences. In other words, even in full-width kanji, hiragana, and symbols, there are characters that are printed unevenly on one side, and according to the conventional method of simply comparing the physical blank width with the printing pitch, excessive space is used. And a problem that a full-width space and a half-width space are erroneously recognized.

[0005] For example, the character string "...
・ An extra half-width or full-width space is inserted after the period in "". Similar phenomena may occur around other parentheses, punctuation, or half-width alphanumeric characters.

When such a phenomenon occurs, the character string output from the character recognition device becomes different from the original character string due to an excessive space or an erroneous recognition of a full-width space and a half-width space. This is a big problem for applications that need to input data to

An object of the present invention is to provide a method of recognizing a space between characters with high accuracy by distinguishing full-width / half-width characters in a character recognition device for recognizing such a Japanese sentence.

[0008]

According to the first aspect of the present invention,
In a character recognition device that processes Japanese sentences, the size of a space between characters is compared with a standard character size in the same line, and the space between the characters is recognized by distinguishing a full-width space and a half-width space. together, if the character recognition results of before or after the inter-character text space portion is a particular character that has been set in advance, and corrects the size of the inter-character space portion, the size of the inter-character space portion after the correction Same as
It is characterized in that space recognition is performed by comparison with a standard character size in a line .

According to a second aspect of the present invention, there is provided a character recognition apparatus for processing a Japanese sentence, wherein a character recognition result of a character before or after a space between characters is a predetermined specific character. The process of correcting the size of the space between characters is performed, and the size of the space between characters or the size of the space between characters after correction is divided by the standard character size in the same line.
The quotient as the number of double-byte spaces, and the remainder
There is a double-byte space for the rest based on the standard character size
Alternatively, space recognition of a half-width space is performed, and the recognition result is added with the number of full-width spaces obtained as the quotient of the division to obtain a final space recognition result.

[0010] The third aspect of the present invention is the first or second aspect.
In the inter-character space recognition method according to the invention described above, a character having a property of being printed near the beginning or end of the line as a specific character is set.

The invention described in claim 4 is the first or second invention.
In the inter-character space recognition method according to the invention described above, half of the standard character size is subtracted by correcting the size of the inter-character space.

[0012] The invention according to claim 5 is the invention according to claim 1 or 2.
In the inter-character space recognition method according to the invention described above, the product of the coefficient depending on the specific character adjacent thereto before or after and the standard character size is subtracted by correcting the size of the inter-character space.

[0013]

Function: For the same line as the space between characters to be processed, the height of the line (width of the line if it is vertical writing) or the width of the character (vertical) except characters that are clearly half-width characters in the character string. If writing, calculate the width (height for vertical writing) of the standard full-width character in the line using the average value of the character height, etc., and calculate this as the standard character size in the line. I do. A method for detecting such a standard character size is known.

It is clear that this standard character size can be used as a reference for the size of a full-width space (character) (width for horizontal writing, height for vertical writing), and that it can also be used as a reference for half-width space. it is obvious.
Therefore, if it is assumed that there is only one full-width space or half-width space between characters, or that there is no space, the size of the space between characters is assumed.
'S (horizontal if the width, if vertical height) compared to the standard character size of the line, em space if the size of the inter-character space portion above the standard character size, smaller than the standard character size, which is half If the size is more than half size space,
Other than that, there is no space, and it is possible to recognize the space between characters. Since the standard character size is a value that more accurately reflects the size of the double-byte space than the print pitch, recognition accuracy can be improved as compared with the method based on the print pitch.

However, such a simple comparison alone gives
Blank spaces between single-byte alphanumeric characters and double-byte characters, and blank spaces between parentheses and punctuation marks printed eccentrically at the beginning or end of the line are misrecognized as spaces, and single-byte spaces are replaced with double-byte spaces. May be mistakenly recognized. Such inconvenience occurs because the size of the space between characters is increased or decreased by the amount that the printing position is unevenly distributed to the beginning or end of the line with respect to the standard printing position of full-width characters,
Correcting the inter-character space size by the increase or decrease and then performing space recognition by comparison with the standard character size allows a correct recognition result to be obtained.

According to the invention of claim 1 and its dependent claims 3, 4 or 5, it is determined whether or not a blank portion between specific characters which causes such an increase or decrease of the blank portion size. Judging from the recognition result, the necessary correction is applied to the space between characters and then compared with the standard character size, so the space between half-width alphanumeric characters, parentheses and punctuation marks is recognized with high accuracy be able to.

There are also various kinds of parentheses and punctuation marks that increase or decrease the size of the space between characters, and each of them has a different increase / decrease value of the space size. According to the fifth aspect of the present invention, since the correction value of the inter-character space size can be optimized according to characters before and after the space, more accurate space recognition can be performed. Become.

According to the invention of claim 2 and the inventions of the dependent claims 3, 4 and 5, the inter-character space size is corrected according to the preceding and following characters, and then divided by the standard character size to obtain the quotient.
Is the number of double-byte spaces, and
The remainder is recognized as a full-width / half-width space based on the standard character size, and the recognition result is obtained as the quotient of the division .
Since the final recognition result is obtained by adding the full-width spaces of the number of characters, high-precision space recognition is possible even when one or more full-width or half-width spaces exist in the inter-character space.

[0019]

FIG. 1 is a schematic block diagram of a character recognition apparatus according to each embodiment of the present invention. In this character recognition device,
The image input unit 11 reads an image of a document by a scanner or the like, inputs the binary image data, and stores the binary image data in the image memory 11 . The line / character cutout unit 12 is provided in the image memory 11.
A character line and a character image are cut out from the input image in the box, the character image data is stored in the character image memory 13, the character cutout position, the character width (here described as horizontal writing, and the character height for vertical writing), The cutout information such as the width of the space between characters (height for vertical writing) and the height of a line (height for vertical writing) is stored in the cutout information memory 14. In addition, a standard character width (height in the case of vertical writing) is calculated for each line using an average value of the line height and character width, and the result is stored in the cutout information memory 14.

The character recognizing section 15 reads out character image data from the character image memory 13, extracts a characteristic amount after normalization, performs matching between the extracted characteristic amount and the dictionary in the character dictionary memory 16 and recognizes a small distance. A result candidate is determined, and its character code and distance data are stored in a recognition result memory 1.
7 is stored.

The space recognizing section 18 refers to the contents of the cut-out information memory 14 and the recognition result memory 17 to perform inter-character space recognition processing, and the number and type of spaces (characters) between characters (separate double-byte / half-width spaces). Is determined, and the information of this space is stored in the recognition result memory 17 in accordance with the character arrangement. Details of the space recognition processing will be described later for each embodiment.

The result output unit 19 outputs the contents of the recognition result memory 17 to a display, a printer, a magnetic disk device, and the like.

[0023] Next, horizontal following string "This is," new BICOH WP is "" including the alphabetic characters of an example when processing, will be explained on the contents of the space recognition process in each example. In the following description, for convenience of notation, a half-width space in a character string is represented by ";" and a full-width space is represented by ":" . According to this notation, the correct notation of the above character string is "This is 'new type; BICOH ; WP'", and the alphabetic characters are half-width. It is also assumed that the character width, character space width, standard character width, and character recognition result as shown in Table 1 are obtained for this character string.

[0024]

[Table 1]

FIG. 2 is a schematic flowchart of the space recognition process, which is common to each embodiment. In this flowchart, processing of a horizontally written document is assumed. In the case of vertical writing, the same processing can be performed by replacing “width” with “height”.

Embodiment 1 Processing is performed in order starting from the space between characters at the head of a line. It is checked whether or not the character recognition result of the character before the focused inter-character space is a specific character for which the space width is to be corrected. Specifically, such a specific character is, at the beginning of the line (left)
Punctuation marks (, .., etc.) and closing parentheses (""｝)> etc.) which are printed to the side.

In the case of such a specific character, a half of the standard character width (here, 30) is subtracted as a correction value from the width of the space between the noticed characters. If it is not a specific character, this correction is not performed.

Next, it is checked whether or not the character after the focused inter-character space is a specific character to be corrected. Specifically, the specific character is an opening parenthesis (“｛(<, etc.). The specific character is a space between the characters to be focused on. From the width of the blank part,
Subtract half the standard character width. If it is not a specific character, this correction is not performed.

Next , two or more spaces exist between the characters.
Therefore, the width of the space between characters after the correction processing is divided by the standard character width . Then, the quotient is stored as the number of double-byte spaces. Note that there are two spaces between characters
If it is assumed that there are no more characters, there is no need to perform this division.

Next, the remainder of the division (the width of the remaining portion excluding the full width space of the space between characters) is calculated as follows:
In addition, based on the comparison based on the standard character width,
The width of the remaining part excluding the space equivalent is
It can be regarded as a half-width space, or space recognition is performed. You
That is, if the remainder value is close to the standard character width, the remainder
(Remainder part) is also recognized as a double-byte space,
If it is up to about half the quasi-character width, it is recognized as a half-width space.
You. Therefore, first, a predetermined value is subtracted from the standard character width and compared with the remainder as a full-width space threshold . If the full-width space threshold is exceeded , the remainder is recognized as one full-width space character. If not a full-width space, then the standard sentence
A value of approximately half of the character width is compared with the remainder as a half-width space threshold, and if it is equal to or larger than the half-width space threshold , the remainder is recognized as one half-width space character. If the remainder is less than the half-width space threshold, it is recognized that there is no space.

That is, here, the standard character width is 60
Yes, for example, if the remainder is 50 or more, double-byte space 1
If the character and the remainder are less than 50 and up to 30 , it is recognized that one half-width space is one character, and if the remainder is less than 30, there is no space.

The final result of the space recognition between the characters of interest is obtained by adding the number of full-width spaces previously obtained to the result of recognition of the remainder.

If it is assumed that two or more spaces do not exist between the characters, the width of the space between characters after the necessary correction according to the preceding and succeeding characters is compared with the full-width space threshold to compare the full-width space threshold. Recognize the space, this is the recognition
If not, a half-width space is recognized by comparing the width of the blank portion itself with the half-width space threshold, and this is the final result.

In the character string exemplified here, the width of the blank portion between the reading point of the fourth character from the head and the next “
It is. However, since the left side is a specific character, half the value 30 of the standard character width 60 is subtracted. Similarly, since the right side is a specific character, 30 is subtracted. As a result, the correction value of the blank portion width is 3 (= 63−30−). 30), the space recognition result is “no space”. The width of the blank portion between the second half and the next character is 35, but since the previous character is a specific character, 30 is subtracted, so that it is determined that there is no space. Table 2 shows the correction result of the width of the blank portion and the space recognition result, and the same character string including the character string and the space in the example is stored in the recognition result memory 1.
7 is obtained.

[0035]

[Table 2]

If the space recognition is performed using the full-width space threshold or the half-width space threshold without performing the correction by the preceding and following characters, the result is as shown in Table 3, and the character string after the processing is "This is:" New type; BICOH ; WP ”;”, and one full-width space and one half-width space will be recognized extra.

[0037]

[Table 3]

Embodiment 2 A correction value is set as shown in Table 4 for each specific character to be corrected for the inter-character space width.

[0039]

[Table 4]

When the character recognition result of the character before and after the space between the characters matches any of the characters in Table 4,
By subtracting the correction value set for the character from the space between characters, the width of the space is corrected, the width after this correction processing is divided by the standard character width, and the full-width / half-width of the remainder is obtained. Space recognition is performed, and a full-width space equal to the quotient is added to the result to obtain a final space recognition result.

In the case of the character string in the example, the blank portion width between the reading point of the third character from the beginning and "is 63, but the correction value of the previous character is 36 (= 60 × 0.6) according to Table 4. From Table 4, the correction value for the succeeding character is 18 (= 60 × 0.3), and the total correction value is 54. Therefore, the blank space width after correction is 9, and thus it is determined that there is no space. Similarly, the result of the space recognition processing is as shown in Table 5, and the space in the character string in the example can be correctly recognized.

[0042]

[Table 5]

In this embodiment, two sentences are also inserted between characters.
If the premise that there is no space equal to or greater than a character is satisfied, the space may be immediately recognized by comparing with the full-width / half-width space width without dividing the corrected inter-character space width.

[0044]

As described above, according to the first to fifth aspects of the present invention, half-size spaces and full-size spaces are mixed, and the size of the space between characters is affected by the uneven distribution of the print positions of the characters before and after. Since the space between characters can be recognized with high accuracy by distinguishing full-width / half-width spaces from a fluctuating Japanese sentence, a character recognition device inputs a character string that is extremely faithful to the original character string including the space. It becomes possible.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram of a character recognition device according to each embodiment of the present invention.

FIG. 2 is a schematic flowchart of a space recognition process.

[Explanation of symbols]

 Reference Signs List 10 Image input unit 11 Image memory 12 Line / character cutout unit 13 Character image memory 14 Cutout information memory 15 Character recognition device 16 Character dictionary memory 17 Recognition result memory 18 Space recognition unit 19 Result output unit

Claims (5)

(57) [Claims] 1. A character recognition apparatus for processing a Japanese sentence, wherein a full-width space and a half-width space are distinguished for an inter-character space by comparing the size of the inter-character space with a standard character size in the same line. Recognize the space, and if the character recognition result of the character before or after the inter-character space is a predetermined specific character, correct the size of the inter-character space and correct the inter-character space. Same as part size
An inter-character space recognition method characterized by performing space recognition by comparing with a standard character size in the same line . 2. A character recognition apparatus for processing a Japanese sentence, wherein when a character recognition result of a character before or after a space between characters is a predetermined specific character, the space between characters is set in advance. The size of the space between characters or the size of the space between characters after correction is divided by the standard character size in the same line , and the quotient is calculated.
Is the number of double-byte spaces, and the remainder is calculated based on the standard character size.
An inter-character space recognition method characterized by performing space recognition of a square space or a half- width space, and adding a full-width space obtained as a quotient of the division to the recognition result to obtain a final space recognition result. 3. The space between characters according to claim 1 or 2.
In recognition method, characterized in that the particular character is a character of the property of being printed closer to beginning side or end of the line side
Intercharacter space recognition method . 4. A space between characters according to claim 1 or 2.
In recognition method, especially the correction of size between character blank portion, that half the value of the standard character size is subtracted
Character recognition method for character space . 5. The space between characters according to claim 1 or 2.
In recognition method, the correction of the size between the character space portion, and wherein the product of the coefficient and the standard character size depending on the particular character adjacent to the front or rear is subtracted
The inter-character space recognition method .