JP2003308317A - Character string conversion method, character recognition method, character recognition device, and character recognition program - Google Patents

Abstract

(57) [Problem] To provide a character recognition device capable of performing correct kana-kanji conversion even when a similar character is erroneously read in recognizing a handwritten character. A second character string that can be converted from a first character string obtained as a result of character recognition and a first character in the first character string are formed into the first character. And a fourth character string that can be converted from the third character string obtained by replacing the third character string with a second character similar to the first character string. Is converted into another character string selected from the output conversion candidates.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device for recognizing handwritten characters.

[0002]

2. Description of the Related Art Conventionally, there is a character recognition device which displays a recognition result as a character code or a character code string from handwriting information about a character or a character string handwritten and input through a coordinate input device such as a coordinate input tablet. To do.

Japanese Patent Application No. 2000-333919 describes a character recognition apparatus and method for recognizing an input character string while appropriately cutting out characters in an environment where character breaks are not explicitly given. .

Further, Japanese Unexamined Patent Publication No. 9-269974 describes a character recognition device for performing character recognition when writing information for each character is obtained.

[0005]

In the character recognition device, it is sometimes impossible to uniquely define similar characters having the same shape. For example, "tsu" and "tsu", "1" and "l", etc. may not be used to know which one is input only by the shape of the input character. Regarding "tsu" and "tsu", the discrimination is usually made based on the size. However, the size in which the characters are written varies depending on the size of the person and the characters before and after, and it is not always possible to distinguish them reliably.

Furthermore, if a kana-kanji conversion is performed on a partially incorrect character recognition result, a completely different character string is obtained.
For example, even if the writer intends to input “Tokyo (Tokyo)”, the result of determining “yo” as “yo” is “toukiyou (for registration)”.

Although such a problem is unlikely to occur in the combination of the keyboard and the kana-kanji conversion, it is more likely to occur in the case of combining the character recognition and the kana-kanji conversion.

The present invention is directed to a character string conversion method and a character recognition method capable of correctly performing the subsequent kana-kanji conversion even when a similar character peculiar to character recognition is subsequently recognized, and a character recognition device and a character using the same. The purpose is to provide a recognition program.

[0009]

The present invention is represented by a coordinate sequence of a pen tip detected by the coordinate input device from the time the pen touches the coordinate input device until the pen leaves the coordinate input device. A character string obtained by performing character recognition based on the stroke sequence is converted into another character string, and another character string that can be converted from the first character string obtained as the character recognition result is converted. It is possible to convert from a second character string and a third character string obtained by replacing the first character in the first character string with a second character having a shape similar to the first character. The other fourth character string is output as a conversion candidate for the first character string, and the first character string is converted to another selected one of the output conversion candidates. It is characterized by

According to the present invention, when recognizing a handwritten character, correct kana-kanji conversion can be performed even if a similar character is misread.

The present invention is based on a stroke sequence represented by a coordinate sequence of a pen tip detected by the coordinate input device between the time the pen touches the coordinate input device and the time the pen leaves the coordinate input device. When character recognition is performed and the first character obtained as a result of character recognition has a second character having a similar shape, the second character is associated with the first character, The other second character string that can be converted from the first character string including the first character obtained as the character recognition result, and the first character in the first character string are converted into the second character string. The other fourth character string that can be converted from the third character string obtained by replacing the first character string is output as the conversion candidate of the first character string, and the first character string is output. It is possible to convert the selected one of the conversion candidates to another character string. The features.

According to the present invention, when recognizing a handwritten character, even if a similar character is misread, kana-kanji conversion can be performed correctly.

[0013] Preferably, when the combination of the third character of the character recognition result immediately before the first character is character-recognized and the second character is a character combination that can be a character string, The second character is associated with the first character.

Further, preferably, when the combination of the third character and the first character is a combination of characters that cannot be a character string, the second character is replaced with the second character.
Is used as the character recognition result, and the conversion candidate of the first character string including the second character is output.

Further, preferably, the first character is a character having a proper use of uppercase and lowercase, and the relationship between the second character and the first character is such that one of them is in uppercase and the other is in the other. Is in a lower case, it is determined whether to associate the second character with the first character based on the size of the written character that is the recognition target of the first character.

In the case of a so-called overwritten character recognizing device, the second character is a combination of two characters written continuously, and the structural relationship between the two characters is the first character. The shape may be similar to the shape.

In the character recognition device of the present invention, the stroke represented by the coordinate series of the pen tip detected by the coordinate input device from when the pen touches the coordinate input device until it leaves the coordinate input device. When a character recognizing unit that performs character recognition based on a string and a first character obtained as a result of character recognition by the character recognizing unit have a second character whose shape is similar, the second character Associating means for associating the above character with the first character, and another second character that can be converted from the first character string including the first character obtained as the character recognition result by the character recognizing means. A string and another fourth character string convertible from a third character string obtained by replacing the first character in the first character string with the second character; Output means for outputting as a character string conversion candidate, The serial first character string, characterized by comprising a conversion means for converting the selected one of the other string of the conversion candidates are output in the output means.

According to the present invention, when recognizing a handwritten character, even if a similar character is misread, kana-kanji conversion can be performed correctly.

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing the functional arrangement of the character recognition apparatus according to this embodiment.

This character recognition device is provided with a tablet 100,
Character recognition unit 101, character recognition dictionary 102, character buffer 103, similar character storage unit 104, similar character addition unit 10
5, kana-kanji conversion unit 107, kana-kanji conversion dictionary 106,
It is composed of the display 108.

The tablet 100 is made of, for example, a transparent plate, and is arranged on the display screen of the display 108 in an overlapping manner. A character writing area 301 (see FIG. 3) is provided on the input surface of the tablet 100. When the user writes a character on the character writing area 301 using the dedicated pen P, the handwriting data (time series information of coordinate values) is detected by the tablet 100 and transferred to the character recognition unit 101. That is, the two-dimensional coordinate data representing the position of the pen tip while the pen P is touching the tablet 100 is sampled at regular time intervals. The obtained coordinate data is sent to the character recognition unit 101.

The character recognizing unit 101 treats a coordinate data string from the time the pen P touches the tablet until it leaves the tablet, that is, a coordinate data string of handwriting, as a group of data called a stroke, and acquires this as stroke data. . Every time the stroke data is input, the character recognition unit 101 calculates from the stroke data input so far,
An optimal character string is recognized using the character recognition dictionary 102.

The character recognition dictionary 102 is composed of, for example, a character structure dictionary and an inter-character structure dictionary.

The character structure dictionary is data (character structure dictionary information) representing the structure of each character to be recognized, that is, for each character, the shape of the strokes that form the character and the positional relationship between the strokes (structure). ) Is a dictionary in which feature information such as is registered.

The inter-character structure dictionary is data (inter-character inter-character) that expresses a structural relationship between two characters written continuously in a plurality of characters registered in the character structure dictionary. This is a dictionary in which dictionary information) is registered.

On the tablet 100, for example, two consecutive characters may be written side by side on the left or right side, or may be overwritten (for example, a character writing area for writing characters with a pen corresponds to one character). There is also only if). The inter-character dictionary information in the former case is characteristic information of the positional relationship (structure) between the stroke of one character and the stroke of the other character when two characters are written side by side on the left and right, and the latter The inter-character dictionary information in this case is characteristic information of the positional relationship (structure) between the stroke of one character and the stroke of the other character when two characters are overwritten and written.

Each time the stroke data is input, the character recognizing unit 101 uses the character structure dictionary and the inter-character structure dictionary to determine the most reliable stroke based on the shape of the strokes input so far and the positional relationship between the strokes. It seems to find the most suitable character (character string). For example, a character code for each character is output as the recognition result.

The character codes output from the character recognition unit 101 are stored in the character buffer 103 in the order of output.

The similar character storage unit 104 collects a plurality of characters (character codes of which are similar in shape) into one similar character group, and a plurality of such similar character groups are stored. There is.

Characters whose shapes are similar to each other are, for example,
As shown in FIG. 2, it is not possible to distinguish which one is input only by the shape of the input characters such as “yo” and “yo”, “1” and “l”, and “tsu” and “tsu”. It is a character like. “Yo” and “yo”, “1” and “l”, “tsu” and “tsu”, etc. are each one similar character group.

The similar character adding unit 105 uses the character buffer 1
For each character stored in 03, it is checked whether the character is included (belongs) in one of the similar character groups stored in the similar character storage unit 104, and if it is included, , The character is included (belongs to)
Each character in the similar character group (characters other than the relevant character)
Is associated with the character in the character buffer 103.

For example, when the character string "shiyo" is stored in the character buffer 103, the character "yo" is a character belonging to one similar character group of the plurality of similar character groups, and is the same. Since the similar character group includes “yo” in addition to “yo”, the similar character adding unit 105
Is the character "yo" stored in the character buffer 103.
The similar character "yo" is stored in association with. Here, this is expressed as “{yo, yo}”, and the character string of the recognition result in the character buffer 103 at this time is “shi {yo, yo}”.
It is written as. Here, “{X, Y}” means that either character (character string) X or character (character string) Y may be used.

The kana-kanji conversion unit 107 uses the character buffer 1
The character string in 03 is analyzed in Japanese, collated with the Kana-Kanji conversion table dictionary 106 stored in advance, and converted into another optimal character string.

In the ordinary kana-kanji conversion, conversion processing is performed on one character string. Here, however, a plurality of character strings are constructed in consideration of a combination of similar characters, and conversion processing is performed for each of them. The results are combined and considered as a change candidate.

For example, the character string "Touki {yo,
If there is "", and kana-kanji conversion is performed on this character string, conversion processing is performed for each of "touki" and "toukiyou."

The conversion candidates for "Tokyo" are "Tokyo",
"Touki", ..., and the conversion candidates for "Toukiyou" are "Registration", "Winter", "Winter", "Toukiyou", and so on, so "Touki {yo, yo} u" conversion The candidates are "Tokyo", "Tokyo", which is a combination of "Tokyo" and "Tokyo" conversion candidates.
"For registration", "for winter", "for winter", "Toukiyou", ...

FIG. 3 shows an appearance of the information equipment as the character recognition device shown in FIG.

As shown in FIG. 3, the user writes a character on the tablet 100 with the pen P on the main surface of the information equipment, that is, the surface on which the transparent tablet 100 is laminated on the display screen of the display 108. Character writing area 30 for
1 and a cursor C indicating the result of recognizing the handwriting written in the character writing area 301 as a character string and the character insertion position.
A recognition result display area 302 for displaying, a Kana-Kanji conversion instruction button 303, a decision button 304 for selecting and confirming a desired conversion candidate from the conversion candidates displayed by the Kana-Kanji conversion processing, and a cursor position A one-character backward delete button 305 for instructing to delete the immediately preceding character is secured.

Next, the processing operation of the character recognition apparatus of FIG. 1 will be described with reference to the flow charts shown in FIGS.

In step S1, when one stroke is written in the character writing area, a coordinate data string of the stroke, that is, stroke data, is sent to the character recognition unit 101.

In step S2, the character recognition unit 101
While referring to the character recognition dictionary 102, the optimum character is recognized from the stroke data corresponding to one character input so far.

In step S3, the optimum character (character code of the character) recognized by the character recognition unit 101 is displayed in the character buffer 1.
03, and the character is displayed on the display 10
8 is displayed.

The similar character adding unit 105 uses the character buffer 1
Every time the character output as a recognition result (code thereof) is stored in 03, the character is included in (belongs to) one of the similar character groups stored in the similar character storage unit 104. Check whether or not (step S
3) If included, a character (code of) similar to the character in the similar character group that includes (belongs to) the character, that is, similar character (character code of)
Is stored in the character buffer 103 in association with the character (step S5). Here, this is "{yo,
"}". Note that this association is not limited to the above method, and there is another character having a similar shape to the character (for example, “yo”) stored as the character recognition result in the character buffer 103, and which character Any method may be used as long as it is possible to understand whether the character is such a character (for example, "yo").

On the other hand, in step S4, when there is no other character whose shape is similar to the character stored in the character buffer 103 (that is, the character stored in the similar character storage unit 104 is similar). If it is not included (does not belong) in any of the character groups, step S
Go to 6.

For example, after the user designates the character string as the character recognition result displayed in the recognition result display area 302 of FIG. 3 as the character string to be converted (or after designating the range of the character string to be converted). 3 is selected, kana-kanji conversion is performed on the character string to be converted (step S6).

When there is a character (first character) to which the similar character is associated in step S5 in the character string to be converted (first character string) (step S in FIG. 5).
7), the process proceeds to step S9 in FIG.
It progresses to step S8 of FIG.

In step S8 of FIG. 5, the kana-kanji conversion unit 107 generates conversion candidates from the first character string to be converted, and in step S9, the kana-kanji conversion unit 107 generates the first conversion candidate.
Conversion of the first character string from the second character string in which the similar character in the first character string and the character (first character) associated with the similar character in the first character string are replaced with the similar character. A candidate is generated (step S9).

Then, the conversion candidates for the first character string generated in step S8 or step S9 are displayed on the display 108 (step S10).

For example, as a recognition result, when "sho" is recognized after the character "shi" and "sho" is stored in the character buffer 103, in step S5,
Since the shapes of “yo” and “yo” are similar, the similar character “yo” is stored in association with the character (code) “yo” as the recognition result stored in the character buffer 103. To do. This is referred to as “{yo, yo}” here. In this case, in step S6, when a kana-kanji conversion instruction is made with the character string "SHO" displayed on the display 108 as the conversion target, the process proceeds from step S7 to step S9, and the character string " It consists of a character string that can be converted from "yo"("shiyo", ...) and a character string that can be converted from the character string "sho"("place","writing", ...). , ...} is generated and displayed on the display 108.

On the other hand, as a recognition result, when “ka” is recognized after the character “shi” and “shika” is stored in the character buffer 103, “shi” or “ka” is stored. Since there is no character having a similar shape (similar characters such as “shi” and “ka” are not stored in the similar character storage unit 105), the character string “shika” is converted in step S6. When a command for kana-kanji conversion is given, the process proceeds from step S7 to step S8, and a character string ({deer, shika, ...}) that can be converted from the character string "shika" is generated as a conversion candidate, and the display 108
Is displayed in.

The user selects a desired conversion candidate from the displayed conversion candidates, and, for example, the "OK" button 3 in FIG.
When 04 is pressed, the first character string is converted into a character string as the selected conversion candidate (step S11).

The above steps S1 to S11 are repeated until the stroke input is completed (step S12).

FIG. 6 shows another configuration example of the character recognition device according to this embodiment. In addition, in FIG.
The same parts as those in FIG. 1 are designated by the same reference numerals, and only different parts will be described. That is, in FIG. 6, the character combination storage unit 109 is further provided.

In the character combination storage unit 109, when a character belonging to any of the plurality of similar character groups stored in the similar character storage unit 104 is combined with another character,
Possible combinations are stored as a character string. For example, in Japanese, "ya", "kyu", "kyo", and "moppora" are used to express the syllabary and consonants.
There are small letters such as "yu", "yo", and "tsu". Characters that are written in small letters to express such whistle sounds and consonants are called "lowercase letters" here, while letters of normal size are called "uppercase letters". Sometimes. When a letter has uppercase and lowercase letters, and there are letters with similar shapes and different sizes, such as "yo" and "yo", that is,
When there are “yo” and “yo” as one similar character group, when “ka” is placed in front of each character as shown in FIG. 7, there may be a combination of characters “kayo”. However, there is no possible combination of the letters "ka". On the other hand, when “shi” is placed before each of the characters “yo” and “yo”, “shiyo” and “sho” are possible character combinations as a character string.

As described above, when each of the characters belonging to any one of the plurality of similar character groups stored in the similar character storage unit 104 is combined with another character (for example, when it is arranged before), the character From the context of
At least one of a combination of characters that can be a character string and a combination of characters that cannot be a character string is stored in the character combination storage unit 109.

For example, here, the character combination storage unit 10
In FIG. 9, for each of the characters belonging to any of the plurality of similar character groups stored in the similar character storage unit 104, another character is arranged immediately before that character as in the upper part of FIG. When the character string is combined with the character, it is assumed that the character string that the combination can be is stored.

In the similar character adding unit 105, the character (first character) stored in the character buffer 103 as the recognition result this time is included in one of the similar character groups stored in the similar character storage unit 104. When (belongs to)
Among the similar characters in the similar character group including (belonging to) the first character, the character of the recognition result immediately before stored in the character buffer 103 (and the similar character is associated with the character). When it is, its similar characters)
The similar character that can be a combination of the characters and the character string (the combination is stored as a character string that can be stored in the character combination storage unit 109) is associated with the first character.

Further, the similar character adding unit 105 determines that the combination of the first character and the character of the recognition result immediately before the first character (and the similar character when the character is associated with the character) is a character string. When it is impossible (the combination is
When not stored as a possible character string in the character combination storage unit 109), the first character in the character buffer 103 is associated with the similar character (there are a plurality of associated similar characters). At that time, any one of them) is rewritten. When there are a plurality of similar characters associated with the first character, the similar characters other than the similar character used for rewriting the first character are associated with the similar character used for the rewriting. To store.

8 to 9 are flow charts for explaining the processing operation of the character recognition device shown in FIG. 8 to 9, the same parts as those in FIGS. 4 to 5 are designated by the same reference numerals, and only different parts will be described. That is, in FIG. 8, step S5a of FIG.
It is replaced by step S5c.

In step S4 of FIG. 8, the similar character adding unit 105 stores the character (first character) obtained as the recognition result of the character recognition unit 101 in the character buffer 103 in step S3. The first character is one of the similar character groups stored in the similar character storage unit 104.
Check whether it is included (belongs to). If it is included, the process proceeds to step S5a, and for each character (similar character) in the similar character group including (or belonging to) the character, the first character stored in the character buffer 103. A combination with a character obtained as a recognition result immediately before is obtained as a recognition result (and a similar character when a similar character is associated with that character) can be a character string (the combination is It is checked whether it is stored as a possible character string in the character combination storage unit 109).

Then, the similar character adding unit 105 determines that, among the similar characters of the first character, the character obtained as the recognition result immediately before the first character is obtained as the recognition result and the character similar to the character. When associated, the combination with each of the similar characters may be a character string (the combination is stored as a possible character string in the character combination storage unit 109). It corresponds to the character of.

In steps S5b to S5c, the similar character adding unit 105 determines that the first character and the character as the recognition result immediately before the first character (and the similar character when the character is associated with the character). Is not possible as a character string (the relevant combination is the character combination storage unit 10
9 is not stored as a possible character string),
The first character in the character buffer 103 is rewritten with a similar character associated with the first character (when there is a plurality of associated similar characters, any one of them is rewritten). When there are a plurality of similar characters associated with the first character, the similar characters other than the similar character used for rewriting the first character are associated with the similar character used for the rewriting. To store.

The processes after step S6 are the same as those described with reference to FIG.

For example, as a recognition result, even though the user intends to input "yo" after the character "shi", it is recognized as "yo" and the character buffer 103 stores
It is assumed that "sho" is stored. In this case, referring to the similar character storage unit 104, “yo” and “yo” have a similar relationship, and further, referring to the character combination storage unit 109,
A character combination of "yo" and the character "shi" of the recognition result immediately before that is possible as a character string. Therefore, in step S5a, the similar character “yo” is stored in association with the character (code) “yo” as the recognition result stored in the character buffer 103. This here
It is written as "{yo, yo}". In addition, the combination of the character "yo" of this recognition result and the character "shi" of the previous recognition result is
Since it can be a character string, step S5b
To step S6. In step S6, when the kana-kanji transformation is instructed with the character string "sho" as the conversion target,
Similar to the case of FIG. 4, the process proceeds from step S7 to step S9, and as the conversion candidates, the character string “shiyo” can be converted (“shiyo”, ...), and the character string “sho” is used. It consists of convertible character strings (“place”, “calligraphy”, ...)
{Place, calligraphy, shiyo, ...} is generated, and the display 10 is displayed.
8 is displayed.

On the other hand, as the recognition result, the user intends to input "yo" after the character "ka",
It is assumed that the character buffer 103 is recognized as “yo” and that “ka” is stored in the character buffer 103. In this case, referring to the similar character storage unit 104, “yo” and “yo” have a similar relationship, and further, referring to the character combination storage unit 109,
A character combination of "yo" and the character "shi" of the recognition result immediately before that is possible as a character string. Therefore, in step S5a, the similar character "yo" is stored in association with the character (code) "yo" as the recognition result stored in the character buffer 103. This here
It is written as “{yo, yo}”. However, the combination of the character "yo" of the recognition result this time and the character "ka" of the previous recognition result cannot be a character string when the character combination storage unit 109 is referred to. Therefore, steps S5b to S5b are performed. Proceed to S5c. In step S5c, the character "yo" of the current recognition result in the character buffer 103 is rewritten with a similar character associated with this character, that is, "yo".
That is, in this case, the character buffer 103 stores the character string “KAYO”. At this time, the character "yo" corrected as the character of this recognition result may be displayed again on the display 108 as the recognition result. Now, in this case, step S6
Then, for example, when the kana-kanji conversion is instructed with the character string “ka” as the currently displayed recognition result as the conversion target, the process proceeds from step S7 to step S8, and the current character buffer 103 is selected as the conversion candidate. Character string that can be converted from the stored character string "Kayo" {Kayo, Kayo,
...} is generated as a conversion candidate, and the display 108
Is displayed in.

As a recognition result, the user inputs "ka" after the character "shi", the character is recognized as "ka", and "shika" is stored in the character buffer 103. . In this case, referring to the similar character storage unit 104, the character “ka” recognized this time does not have a similar character. Therefore, the process directly proceeds from step S4 to step S6. In step S6, when the character string “shika” is designated as the conversion target, the process proceeds from step S7 to step S8, and as the conversion candidates, the character strings that can be converted from the character string “shika” are {deer, shika, ...}. It is generated as a conversion candidate and displayed on the display 108.

As described above, according to the above-described embodiment, when a character in the first character string obtained as a character recognition result has a similar character that may be erroneously recognized, the first character string is detected. Other second character string that can be converted from the character string of No. 3 and the character in the first character string is replaced with the similar character
The other fourth character string that can be converted from the character string of is output as a conversion candidate of the first character string, and the first character string is selected from the output conversion candidates. By converting into one character string, when recognizing a handwritten character, even if a similar character is misread, kana-kanji conversion can be performed correctly.

In the above embodiment, the case of Kana-Kanji conversion has been described as an example, but the present invention is not limited to this case, and can be applied to the case of converting to Katakana, for example.

Further, as the number of similar characters associated with the character of the recognition result increases, the processing amount of generation of conversion candidates also increases, and the number of conversion candidates also increases, and after the user gives a conversion instruction. The response until the conversion candidate is displayed becomes slow, and the time and effort for the user to search for a desired one from many conversion candidates increases. Therefore, when associating similar characters, it is desirable to eliminate as many unnecessary similar characters as possible at that time.

In particular, it is easy to be erroneously recognized because there are characters with different uppercase and lowercase letters (for example, "ya" and "ya",
Therefore, there are many opportunities to associate similar characters in the processing of the similar character adding unit 105.
It is such a character.

Therefore, the similar character adding unit 105 of the character recognition device shown in FIGS. 1 and 6 is similar to the recognized character stored in the character buffer 103 based on the size of the character written on the tablet 100. It may be possible to determine whether or not to associate the characters.

For example, in step S5 of FIG. 4 and step S5a of FIG. 8, the similar character adding unit 105 replaces the similar character stored in the similar character storage unit 104 with the character of the recognition result stored in the character buffer 103. When associating
When the similar character has a relationship between the character of the recognition result stored in the character buffer 103 and the uppercase letter and the lowercase letter, or conversely, the lowercase letter and the uppercase letter, whether or not to associate the similar letter on the tablet 100. The determination may be made based on the size of the written handwritten character (written character) to be recognized.

Whether or not the recognition result character stored in the character buffer 103 and its similar character are in uppercase and lowercase or lowercase and uppercase can be determined from those character codes. Therefore, the similar character storage unit 10
In 4, it is necessary to store the correspondence between the uppercase character code and the lowercase character code in advance. In addition, the similar character storage unit 104 acquires information regarding the size of a written character (written character) from the tablet 100.

The similar character adding unit 105 uses the character buffer 1
When the character of the recognition result stored in 03 is a capital letter (for example, when it is "yo"), the size of the handwritten character is determined in advance from the information about the size of the handwritten character obtained from the tablet 100. When it is less than or equal to the first reference value or smaller than the first reference value, a lowercase letter (in this case, "yo") is stored in the character buffer 103 as a similar character to the character of the recognition result. In addition, when the character of the recognition result stored in the character buffer 103 is a lower case character (for example, “ya”), the size of the handwritten character is calculated from the information on the size of the handwritten character obtained from the tablet 100. When it is equal to or larger than the second reference value set in advance or larger than the second reference value, an uppercase character (in this case, "or") is stored in the character buffer 103 as a similar character to the character of the recognition result.

The first reference value and the second reference value may be predetermined based on the size of the character writing area 301.

Therefore, even if the character of the recognition result stored in the character buffer 103 is an uppercase character, the size of the written character that is the recognition target is larger than the first reference value or the first reference value. In the above case, since it can be said that it is an uppercase letter with high probability, the lowercase letter of the character is not associated as a similar character. Also, the character buffer 10
Even if the character of the recognition result stored in 3 is a lowercase letter, if the size of the written character that is the recognition target is smaller than the second reference value or less than the second reference value,
Since it can be said that it is a small letter with a high probability, the upper case of the character is not associated as a similar character.

In this way, when associating the similar character with the character of the recognition result in which uppercase letters and lowercase letters are properly used, the character of the recognition result to which the similar character is associated is determined by judging based on the size of the written character. The number can be minimized.

Further, when the character writing area is only for one character and the character recognizing unit 101 performs character recognition from the result of writing by writing two consecutive characters on the character writing area (that is, the user is aware of character breaks). The character recognition unit 10 can simply write characters continuously in the character writing area without writing.
In the case of a so-called overwriting character recognition device in which 1 can automatically recognize the handwriting as a character string), the delimiter between characters may not be accurate. For example, even if “mo” is obtained as the recognition result, it may be “shi” and “ko” or “ko” and “shi”.
Therefore, in the case of such an overwriting character recognition device,
It can be said that "mo" and "shi""ko" or "ko""shi" have similar relations. That is, in the case of such an overwriting character recognition device, a combination of two characters written in succession, such as "shi", "ko", or "ko", "shi", Even if the shape of a character having a structural relationship between them is similar to that of a character (in this case, also “”), it is necessary to store such two characters together with the character as a similar character group in the similar character storage unit 104. Other than this, it is the same as described above.

That is, it is a combination of two characters written in succession to the first character (for example, "mo") obtained as a result of character recognition, and the structural relationship between the two characters is the first. A second character that resembles the shape of the character (for example, "shi")
")" Is present, the second character is replaced by the first character.
Other second character string that can be converted from the first character string including the first character obtained as a result of character recognition and the first character in the first character string. And the other fourth character string that can be converted from the third character string obtained by replacing the character of the above with the second character as a conversion candidate of the above-mentioned first character string.

At this time, when the combination of the third character of the character recognition result immediately before the first character is character-recognized and the second character is a character combination that can be a character string, The second character is associated with the first character.

Further, when the combination of the third character of the character recognition result immediately before the first character is character-recognized and the first character is a character combination that cannot be a character string,
The second character, which is a combination of two characters, is used as the character recognition result instead of the first character, and the conversion candidate of the first character string including the second character is output.

[0083]

As described above, according to the present invention, when recognizing a handwritten character, even if a similar character is misread, kana-kanji conversion can be performed correctly.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration example of a character recognition device according to an embodiment of the present invention.

FIG. 2 is a diagram for explaining a similar character group stored in a similar character storage unit.

FIG. 3 is a diagram showing an appearance of an information device as the character recognition device shown in FIG.

4 is a flowchart for explaining a processing operation of the character recognition device shown in FIG.

5 is a flowchart for explaining a processing operation of the character recognition device shown in FIG.

FIG. 6 is a diagram showing another configuration example of the character recognition device.

FIG. 7 is a diagram for explaining a combination of characters that can be a character string and a combination of characters that cannot be a character string.

8 is a flowchart for explaining the processing operation of the character recognition device shown in FIG.

9 is a flowchart for explaining the processing operation of the character recognition device shown in FIG.

[Explanation of symbols]

100 ... Tablet 101 ... Character recognition unit 102 ... Character recognition dictionary 103 ... Character buffer 104 ... Similar character storage unit 105 ... Similar character addition section 106 ... Kana-Kanji conversion dictionary 107 ... Kana-Kanji conversion unit 108 ... Display 109 ... Character combination storage unit

Continued front page    F-term (reference) 5B009 KA01 LA02                 5B064 AB04 BA06 DD08 DD11 EA10                       EA18 FA04 FA06 FA13

Claims (13)

[Claims] 1. A character recognition based on a stroke sequence represented by a coordinate sequence of a pen tip detected by the coordinate input device from when the pen touches the coordinate input device until it leaves the coordinate input device. Is a character string conversion method for converting a character string obtained by performing the above to another character string, and another second character string that can be converted from the first character string obtained as a character recognition result. And another fourth character convertible from the third character string obtained by replacing the first character in the first character string with the second character having a shape similar to the first character. And outputting a character string as a conversion candidate for the first character string, and converting the first character string into another selected one of the output conversion candidates. String conversion method. 2. A character recognition based on a stroke sequence represented by a coordinate series of a pen tip detected by the coordinate input device from when the pen touches the coordinate input device until it leaves the coordinate input device. When the first character obtained as a result of character recognition has a second character whose shape is similar, the second character is associated with the first character, and the character Another second character string that can be converted from the first character string including the first character obtained as a recognition result, and the first character in the first character string are converted into the second character. Is output as a conversion candidate of the first character string and another fourth character string that can be converted from the third character string obtained by replacing the first character string with the output conversion candidate. Characterized by converting to one of the other selected strings of Character recognition method that. 3. When the combination of the third character of the character recognition result immediately before character recognition of the first character and the second character is a combination of characters that can be a character string, the second character The character recognition method according to claim 2, wherein the character is associated with the first character. 4. When the combination of the third character of the character recognition result immediately before the first character is character-recognized and the first character is a combination of characters that cannot be a character string, The character recognition method according to claim 2, wherein the second character is used as a character recognition result in place of the first character, and a conversion candidate of the first character string including the second character is output. . 5. The first character is a character having different upper and lower case letters, and the relationship between the second character and the first character is that one of them is uppercase and the other is lowercase. When there is a relationship, the second character is added to the first character based on the size of the written character that is the recognition target of the first character.
4. The character recognition method according to claim 3, wherein it is determined whether or not the character is associated. 6. The second character is a combination of two characters written in succession, and the structural relationship between the two characters is similar to the shape of the first character. The character recognition method according to item 2. 7. A character recognition based on a stroke sequence represented by a coordinate series of a pen tip detected by the coordinate input device from when the pen touches the coordinate input device until it leaves the coordinate input device. And a first character obtained as a result of character recognition by the character recognition means.
When a second character having a similar shape is present in the character of, the result obtained as the character recognition result by the associating unit that associates the second character with the first character and the character recognizing unit. Another second that can be converted from the first character string including the first character
And another fourth character string convertible from the third character string obtained by replacing the first character in the first character string with the second character. An output unit that outputs the first character string as a conversion candidate, and a conversion unit that converts the first character string into another selected one of the conversion candidates output by the output unit. A character recognition device characterized by being provided. 8. A character recognition based on a stroke sequence represented by a coordinate series of a pen tip detected by the coordinate input device between the time the pen touches the coordinate input device and the time the pen leaves the coordinate input device. And a storage unit for storing a plurality of such similar character groups, the storage unit storing a plurality of such similar character groups. First obtained as a character recognition result
Is a character belonging to a first similar character group that is one of the plurality of similar character groups, the first character is other than the first character belonging to the first similar character group. Associated means for associating the second character of the second character with another second character that can be converted from the first character string including the first character obtained as a result of character recognition by the character recognition means.
And another fourth character string convertible from the third character string obtained by replacing the first character in the first character string with the second character. An output unit that outputs the first character string as a conversion candidate, and a conversion unit that converts the first character string into another selected one of the conversion candidates output by the output unit. A character recognition device characterized by being provided. 9. The associating means is a combination of characters in which a combination of a third character of the character recognition result immediately before the first character is recognized and the second character is a character string. 9. The character recognition device according to claim 7, wherein the second character is associated with the first character at a certain time. 10. When the combination of the third character of the character recognition result immediately before the first character is character-recognized and the first character is a combination of characters that cannot be a character string,
9. The conversion candidate of the first character string including the second character is output instead of the first character as the character recognition result, and the conversion candidate is output. Character recognizer. 11. The first character is a character that has different uppercase and lowercase letters, and the relationship between the second character and the first character is that one of them is uppercase and the other is lowercase. 10. When there is a relation, it is determined whether or not to associate the second character with the first character based on the size of the written character that is the recognition target of the first character. Character recognition device described. 12. The second character is continuously written 2
9. The character recognition device according to claim 7, wherein the character recognition device is a combination of characters and the structural relationship between the two characters is similar to the shape of the first character. 13. A character recognition program for performing character recognition based on a stroke string representing a character written by a pen on a coordinate input device, wherein the computer recognizes a first character string obtained as a character recognition result. From another second character string that can be converted from a third character obtained by replacing the first character in the first character string with a second character that is similar in shape to the first character. Outputting another fourth character string that can be converted from a character string as a conversion candidate for the first character string, and selecting the first character string from the selected conversion candidates of the output conversion candidates. A character recognition program that executes the steps of converting into one conversion candidate and.