JPH08329073A - Handwriting input character processing method - Google Patents

Abstract

PURPOSE: To increase the recognition rate of handwritten characters, improve the efficiency of subsequent conversion processing, and shorten the processing time by determining candidate output order in character string units. CONSTITUTION: The handwritten character string is recognized, and processed character by character, and the recognition result is displayed (Al-A4). Here, priority decision processing is performed for all recognition candidates for the respective characters obtained by the recognition processing and the candidates are rearranged in output order and redisplayed so that candidates which are adequate for the character string are given (A5-A10). According to the output order obtained by this priority decision processing, Japanese syllabary-Chinese character conversion processing is performed and its conversion result is displayed (A12 and A13).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document preparation device such as a Japanese word processor, and more particularly to a handwriting input character processing method used in a document preparation device capable of handwriting input by pen input or the like.

[0002]

2. Description of the Related Art Conventionally, there are some document creating apparatuses such as Japanese word processors that directly input characters by handwriting on a display by pen touch operation. In such a handwriting input type document creating apparatus, a character input by handwriting is recognized and kana-kanji conversion is performed based on the recognition result.

In the handwriting character recognition process, a recognition candidate is usually obtained from the character recognition dictionary by comparing an input character (handwritten character) with a standard character in the character recognition dictionary. In this case, when a plurality of recognition candidates are obtained, the output rank is determined according to the degree of matching with the input character. That is, the character with the highest degree of matching is set as the first candidate and is preferentially output. After that, other candidate characters are output in descending order of matching degree by the user's next candidate selection operation.

[0004]

As described above, conventionally, when a plurality of recognition candidates are obtained by the character recognition process, the output order of the recognition candidates is set for each character, and this is set as the final recognition result. Was output as. For this reason, for example, when a character string such as "star" is input by handwriting,
The recognition candidates for each character are output in units of "s", "t", "a", and "r". Therefore, for example, if the second character "t" is erroneously recognized as "chi", even if "t" is listed as the second candidate, it cannot be saved and "a chi ar" is not saved. A recognition result that does not exist as a character string such as is output.

As described above, even when inputting as a character string, the candidate output order is determined on a character-by-character basis. Therefore, even if there is a valid candidate after the first candidate when viewed as a character string, Could not be saved, and a recognition result that did not exist as a character string was output. For this reason, there is a problem that the subsequent kana-kanji conversion (including kana-kanji conversion kana-kanji conversion) is affected, resulting in a long processing time.

The present invention has been made in view of the above points, and by determining the candidate output order for each character string, the recognition rate of handwritten characters is increased, and the efficiency of the conversion processing performed later is improved and the processing is performed. An object is to provide a handwritten input character processing method capable of improving time.

[0007]

According to the handwritten input character processing method of the present invention, character recognition processing is performed on character string information input by handwriting on a character-by-character basis, and then a recognition candidate for each character obtained by the character recognition processing. (1) Priority determination process regarding character arrangement, (2) Priority determination process regarding phonetic reading element, (3) Priority determination process regarding reverse phonetic reading element, (4) Priority determination process regarding learning, (5) A priority determination process for dictionary lookup, (6) priority determination process for grammar, (7) priority determination process for semantic process, (8) priority determination process for difficult characters, or (9) priority determination process for selection of recognition result. By doing so, the output order of the character recognition result is set, and thereafter, the kana-kanji conversion processing or the kana-kanji conversion processing mixed with kanji is performed according to the output order. And butterflies.

[0008]

According to the above method, the output order is set such that the combination of candidates that are valid as a character string is given priority among the combinations of the recognition candidates for each character obtained by the character recognition processing.

That is, in (1) the priority discrimination processing relating to the character arrangement, the candidates according to the character arrangement information stored in the priority discrimination dictionary in the combination of the recognition candidates for each character obtained by the character recognition processing. It is checked whether or not there is, and if there is a corresponding candidate, the output order of the character recognition result is set based on the candidate.

Further, (2) in the priority discrimination processing concerning the phonetic reading elements, the phonetic reading elements stored in the phonetic reading element table are included in the combination of the recognition candidates for each character obtained by the character recognition processing, and It is checked whether or not there is a candidate having a delimiter code that delimits a phrase in the middle of the phonetic reading element, and if there is such a candidate, the output order of the character recognition result is set based on the candidate.

Further, (3) in the priority discrimination processing concerning the reverse reading element, there is a candidate having the reading element stored in the reverse reading element table in the combination of the recognition candidates for each character obtained by the character recognition processing. Whether or not there is a corresponding candidate, the output order of the character recognition result is set based on the candidate.

Further, (4) in the priority determination process for learning, it is checked whether or not there is a candidate that matches the learning contents of various learning dictionaries in the combination of recognition candidates for each character obtained by the character recognition process. If there is a corresponding candidate, the output rank of the character recognition result is set based on the candidate.

Further, (5) in the priority determination process relating to the dictionary lookup, an extended dictionary and various new word dictionaries for kana-kanji conversion or kana-kanji kana conversion are included in the combination of recognition candidates for each character obtained by the character recognition process. It is checked whether or not there is a candidate that matches the registered content, and if there is a corresponding candidate, the output order of the character recognition result is set based on the candidate.

(6) In the grammar priority determination process, there is a candidate according to the grammar rule in the kana-kanji conversion or kana-kanji kana-kanji conversion process in the combination of recognition candidates for each character obtained by the character recognition process. Whether or not there is a corresponding candidate, the output order of the character recognition result is set based on the candidate.

Further, (7) in the priority determination process relating to the semantic process, a rule of the semantic process in the kana-kanji conversion or kana-kanji kanji conversion process in the combination of recognition candidates for each character obtained by the character recognition process, or
Whether or not there is a candidate according to the learning information of the semantic processing is checked, and if there is a corresponding candidate, the output order of the character recognition result is set based on the candidate.

Further, (8) in the priority discrimination processing for difficult characters, according to the priority information of difficult-to-input characters stored in the priority discrimination dictionary in the combination of recognition candidates for each character obtained by the character recognition processing. It is checked whether there is a candidate, and if there is a corresponding candidate, the output order of the character recognition result is set based on the candidate.

Further, (9) in the priority discrimination processing regarding the selection of the recognition result, it is examined whether or not there is a fixed character in the combination of the recognition candidates for each character obtained by the character recognition processing, If there is a candidate that does, the candidate is given the highest priority, and the output order of the character recognition result is set for the subsequent candidates.

By such a priority determination process, it becomes possible to preferentially obtain a proper candidate as a character string, and as a result, the recognition rate can be improved. Further, the kana-kanji conversion process or the kana-kanji mixed kanji characters after that can be obtained. The conversion rate in the conversion process can also be improved and the overall processing time can be shortened.

[0019]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a document creation device having a kana-kanji conversion function. This device is provided with a CPU 11,
Storage device (1) 12, storage device (2) 13, input / output control unit 14, pen 15, tablet 16, LCD (Liquid C)
crystal display) 17.

The CPU 11 controls the entire apparatus, and here the two CPUs 1 are used for the purpose of speeding up.
It has 1a and 11b. The main CPU 11a is used for input / output processing. The sub CPU 11b is used for recognition processing.

The storage device (1) 12 is composed of, for example, a ROM, and here, the character recognition means 12a and the character recognition dictionary 12 are used.
b, kana-kanji conversion means 12c, kana-kanji conversion dictionary 12
d, a priority determination means 12e, a priority determination dictionary 12f, a phonetic reading element table 12g, a reverse phonetic reading table 12h, and a kana-kanji conversion expansion dictionary 12i.

The character recognition means 12a is means (program) for performing character recognition processing on the input handwritten character by referring to the character recognition dictionary 12b. Character recognition dictionary 12
Reference character b is a standard dictionary for character recognition, which stores information such as standard character patterns.

The kana-kanji conversion means 12c is means (program) for converting a character-recognized character or a key-input character into a kanji character. Kana-Kanji conversion dictionary 12d
Is a standard dictionary for kana-kanji conversion, and stores information such as readings and headings (conversion words) corresponding to the readings.

The priority determining means 12e is means (program) for determining the output order of the candidate characters obtained by the character recognition processing. Specifically, the order of dictionary lookup for kana-kanji conversion, the order of kana-kanji conversion processing, the order of displaying character recognition results, the determination of character candidates not to be processed in the future, and the like are performed.

The priority discrimination dictionary 12f stores information for performing the priority discrimination processing (see FIG. 12). The phonetic reading element table 12g stores information indicating a sequence of characters that are less likely to be a segment break (see FIG. 8).

The reverse reading table 12h stores information indicating a sequence of characters that cannot be used as words in most cases (see FIG. 9). Kana-Kanji conversion extended dictionary 12i
Is an extended dictionary for Kana-Kanji conversion, and is a dictionary for Kana-Kanji conversion prepared separately from a standard dictionary such as a field-specific dictionary or a user-specific dictionary.

The storage device (2) 13 is composed of, for example, a RAM, and here is a character recognition work buffer 13a, a character recognition output buffer 13b, a kana-kanji conversion processing work buffer 13c, a kana-kanji conversion output buffer 13d, and a priority discrimination processing work. Buffer 13e, output buffer 13f for priority determination processing, various learning dictionaries 13g, various new word dictionaries 1
Has 3h.

The character recognition work buffer 13a is a work area used in the character recognition process. The character recognition output buffer 13b is an area for storing a character recognition result (recognition candidate obtained by the character recognition process).

Kana-Kanji conversion processing work buffer 13c
Is a work area used during the Kana-Kanji conversion process.
The kana-kanji conversion output buffer 13d is an area for storing kana-kanji conversion results (conversion candidates obtained by kana-kanji conversion processing).

The work buffer 13e for priority discrimination processing is
This is a work area used in the priority determination process. The priority determination processing output buffer 13f is an area for storing a priority determination processing result (a recognition candidate obtained by the priority determination processing).

As the various learning dictionaries 13g, for example, "provisional learning dictionary" (see FIG. 3), "long-term learning dictionary" (see FIG. 4), "single-kanji learning dictionary" (see FIG. 5), "automatic co-occurrence learning""Dictionary" (see FIG. 6), "bunsetsu learning dictionary" (see FIG. 7), "neuro activation value table" (see FIG. 10),
There are "one character learning dictionary for character recognition" (see FIG. 16) and "character string learning dictionary for character recognition" (see FIG. 18).

The various new word dictionaries 13h are, for example, "Kana-Kanji conversion new word dictionary" (see FIG. 13), "Character recognition single character new word dictionary" (see FIG. 15), "Character recognition character string new word dictionary" (see FIG. 17). I have a dictionary.

The input / output control unit 14 controls input of keys and handwritten characters and screen display. Pen 15
Is used as one of the input devices for inputting handwritten characters and the like to the tablet 16.

The tablet 16 is a coordinate input device,
It is a target for pen input (see FIG. 11). By writing the pen 15 on the tablet 16, handwritten characters can be input.

The LCD 17 is a display device for displaying the recognition result and the like, and is integrated with the tablet 16 here. FIG. 2 is a block diagram showing the configuration of a document creating apparatus having a kanji / kanji conversion function. This device includes a CPU 21, a storage device (1) 22, and a storage device (2).
23, an input / output control unit 24, a pen 25, a tablet 26,
It has an LCD (Liquid Crystal Display) 27.

The CPU 21 controls the entire apparatus. Here, two CPUs 2 are used for the purpose of speeding up.
It has 1a and 21b. The main CPU 21a is used for input / output processing. The sub CPU 21b is used for recognition processing.

The storage device (1) 22 comprises, for example, a ROM, and in this case, the character recognition means 22a and the character recognition dictionary 22.
b, a kanji-mixed kana-kanji conversion means 22c, a kanji-mixed kana-kanji conversion dictionary 22d, a priority determination means 22e, a priority determination dictionary 22f, a phonetic reading element table 22g, a reverse reading table 22h, and a kanji-mixed kana-kanji conversion expanded dictionary 22i.

The character recognition means 22a is means (program) for performing character recognition processing on the input handwritten character by referring to the character recognition dictionary 22b. Character recognition dictionary 22
Reference character b is a standard dictionary for character recognition, which stores information such as standard character patterns.

The kanji-mixed kana-kanji conversion means 22c is means (program) for converting a character-recognized character or a key-input character into a kanji character. The kanji-mixed kana-kanji conversion dictionary 22d is a standard dictionary for kanji-mixed kana-kanji conversion, and stores information such as kanji (converted words) corresponding to the reading of a character string containing kanji.

The priority determining means 22e is means (program) for determining the output rank of the candidate characters obtained by the character recognition processing. Specifically, the order of dictionary lookup for kanji-kana-kana conversion, the order of kanji-kana-kanji conversion processing, the order of displaying character recognition results, and the determination of character candidates not to be processed in the future are performed.

The priority determination dictionary 22f stores information for performing priority determination processing (see FIG. 12). The phonetic reading element table 22g stores information indicating a sequence of characters that are less likely to be a segment break (see FIG. 8).

The reverse sound reading table 22h stores information indicating a sequence of characters that is not possible as a word in most cases (see FIG. 9). The kana-kanji-kanji conversion expansion dictionary 22i containing kanji is an expansion dictionary for kana-kanji conversion, and is a kana-kanji conversion dictionary prepared separately from a standard dictionary such as a field-specific dictionary or a user-specific dictionary.

The storage device (2) 23 comprises, for example, a RAM. Here, a character recognition work buffer 23a, a character recognition output buffer 23b, a kanji mixed kana / kanji conversion processing work buffer 23c, a kanji mixed kana / kanji conversion output buffer 23d, and a priority. Work buffer 23 for discrimination processing
e, an output buffer 23f for priority determination processing, various learning dictionaries 23g, and various new word dictionaries 23h.

The character recognition work buffer 23a is a work area used in the character recognition process. The character recognition output buffer 23b is an area for storing character recognition results (recognition candidates obtained by the character recognition process).

The kanji / kanji / kanji conversion processing work buffer 23c is a work area used during kanji / kanji / kanji conversion processing. The kanji and kana-kanji conversion output buffer 23d is an area for storing kanji and kana-kanji conversion results (conversion candidates obtained by kana-kanji conversion processing).

The work buffer 23e for priority discrimination processing is
This is a work area used in the priority determination process. The priority determination processing output buffer 23f is an area for storing a priority determination processing result (a recognition candidate obtained by the priority determination processing).

As various learning dictionaries 23g, for example, "provisional learning dictionary" (see FIG. 3), "long-term learning dictionary" (see FIG. 4), "single-kanji learning dictionary" (see FIG. 5), "automatic co-occurrence learning""Dictionary" (see FIG. 6), "bunsetsu learning dictionary" (see FIG. 7), "neuro activation value table" (see FIG. 10),
There are "one character learning dictionary for character recognition" (see FIG. 16) and "character string learning dictionary for character recognition" (see FIG. 18).

The various new word dictionaries 23h are, for example, "Kanji mixed kana-kanji conversion new word dictionary" (see FIG. 14), "character recognition single character new word dictionary" (see FIG. 15), and "character recognition character string new word dictionary" (Fig. There is a dictionary.

The input / output control unit 24 controls the input of keys and handwritten characters and the screen display. Pen 25
Is used as one of the input devices for inputting handwritten characters and the like into the tablet 26.

The tablet 26 is a coordinate input device,
It is a target for pen input (see FIG. 11). By writing the pen 25 on the tablet 26, handwritten characters can be input.

The LCD 27 is a display device for displaying the recognition result and the like, and is integrated with the tablet 26 here. FIG. 3 is a diagram showing the structure of the provisional learning dictionary.
The provisional learning dictionary is a dictionary in which the word selected at the time of document creation is provisionally learned. In this example, two words “Kanji” and “Tokyo” are provisionally learned.

In FIG. 3, reference numeral 31 is a dictionary number of "Kanji". Reference numeral 32 is a dictionary number of "Tokyo". Reference numeral 33 is an end mark, which means the end of the provisional learning information.

FIG. 4 is a diagram showing the structure of the long-term learning dictionary. The provisional learning dictionary is a dictionary in which a word selected at the time of creating a document is learned for a long time. In this example, the case where two words "alpha" and "golfer" are learned for a long time is shown.

In FIG. 4, reference numeral 41 is the number of learning times of the word "alpha", and since the dictionary number of "alpha" is "0000", the index "00" of the long-term learning dictionary.
This learning count is stored in the position "00". The number of times of learning is 5.

Reference numeral 42 indicates the number of times of learning of the word "golfer", and the dictionary number of "golfer" is "00".
01 ", so long-term learning dictionary index" 0001 "
This learning count is stored in the position of. The number of times of learning is zero.

FIG. 5 is a diagram showing the structure of a single-kanji learning dictionary. The single-kanji learning dictionary is a dictionary in which a single kanji (one kanji) selected when creating a document is learned. In this example,
It shows a case where two words, "heaven" and "feather", have been learned by single kanji.

In FIG. 5, reference numeral 51 is learning information of a word for which the first single-kanji learning is performed, and in this case, the dictionary number of the single-kanji dictionary of "Ten" is written. Reference numeral 52 is learning information of a word for which the second single Kanji learning has been performed, and in this case, the dictionary number of the single Kanji dictionary of "ha" is written. Reference numeral 53 is an end mark, which means the end position of the single-kanji learning information.

FIG. 6 is a diagram showing the structure of the automatic co-occurrence learning dictionary. An automatic co-occurrence learning dictionary is a dictionary that learns words that are in a co-occurrence relationship. This example is "Apply for Sweepstakes"
And the case where the words that exist in the phrase "Fairy encounter" are automatically co-occurring learned.

In FIG. 6, reference numeral 61 is a front side table, which is a table for storing the information of the front side word by the combination of the words that have been subjected to the automatic co-occurrence learning. Reference numeral 62 is a rear side table, which is a table for storing information of the rear side words by the combination of the words that are subjected to the automatic co-occurrence learning. Reference numeral 63
Is an adjunct and is an adjunct between words that have been auto-co-occurringly learned.

FIG. 7 is a diagram showing the structure of the phrase learning dictionary. The bunsetsu learning dictionary is a dictionary in which bunsetsu breaks are learned. In this example, it is a dictionary in which two phrases, "I wipe a kimono" and "I made it", are learned.

In FIG. 7, reference numeral 71 is learning information for the first phrase learning. Reference numeral 71.1 is the reading of the clause on the front side. The last reading of this clause ("ha" in this example)
Set up the MSB. Reference numeral 71.2 is the reading of the clause on the rear side. The MSB is set for the last reading ("gu" in this example) of this clause. Reference numeral 72 is learning information for the second phrase learning.

FIG. 8 is a diagram showing the structure of the phonetic reading element table. A phonetic reading element is a pattern that is less likely to separate bunsetsu at that position in the reading sequence. For example, in the case of inputting "Shumatsu Matsumoto", the sequence of characters "Shu" and "Shu" It is unlikely that the bunsetsu will be separated. Such a pattern of letters is called a phonetic reading element. In this example, "cha", "chan",
"Chau", "sha", "shan", and "shu" are the reading elements. Candidates whose phrases are separated in the middle of the phonetic reading element are not output, or the output priority is lowered.

In FIG. 8, reference numeral 81 is the reading of the first character, and the first character of the phonetic reading element is registered. Reference numeral 82 is the second character reading, and the second character of the phonetic reading element is registered. Reference numeral 83 is the reading of the third character, and the third character of the phonetic reading element is registered.

FIG. 9 is a diagram showing the structure of the reverse sound reading element table. The reverse reading element indicates a sequence of characters that is unlikely to be a word in most cases. For example, since there is no word that contains a sequence of characters such as "hue",
If there is a character recognition candidate "shu" and a character recognition candidate "shue", "shu" is the first candidate. It is also used to judge the breaks in the kana-kanji conversion clauses and the range of unknown words.

In FIG. 9, reference numeral 91 is the reading of the first character, and the first character of the reverse sound reading element is registered. Reference numeral 92 is the reading of the second character, and the second character of the reverse sound reading element is registered. Reference numeral 93 is the reading of the third character, and the third character of the reverse sound reading element is registered.

FIG. 10 is a diagram showing the structure of the neuroactivity value table. The neuro-activity value indicates the strength of connection between words. In this example, "secretary",
The activity value of each word such as “banquet”, “tavern”, “bar”, and “party” is stored.

In FIG. 10, reference numeral 101 is a neuro number, which is an identification number sequentially attached to each word in which the neuro net is set. Reference numeral 102 is an activation value, which is a value calculated by a neuro calculation formula when the word is fixed.

FIG. 11 is a diagram showing the structure of the tablet. Here, as shown in FIG. 1 or FIG. 2, it is integrated with an LCD, and is configured to point on the display screen of the LCD with a pen.

In FIG. 11, reference numeral 111 is a document edit screen, which is a screen for displaying a document being created or modified. Reference numeral 112 is an operation window, and characters or keys are input by pointing or dragging in this area.

Reference numeral 113 is a handwritten character input frame into which a handwritten character is input. Reference numeral 114 is a single Chinese character key, and the character input by pointing this key is converted to single Chinese character. Reference numeral 115 is a selection / execution key, which is a key for confirming a candidate converted into kana-kanji.

Reference numeral 116 denotes a conversion / next candidate key, which is pointed at when the input character is converted into kana-kanji or when the next candidate of the character already converted into kana-kanji is displayed. Reference numeral 117 is a non-conversion key, which is pointed when the input character is not converted.

FIG. 12 is a diagram showing the structure of the priority discrimination dictionary. This priority determination dictionary is a dictionary used in the priority determination processing, and stores information indicating a difficult character to be prioritized, a character, or a combination of character types.

In FIG. 12, reference numeral 121 is a hard-character block. By storing words that are expected to be written in Hiragana for reasons such as a large number of strokes, easy to forget, and difficult to write (convex, lag, etc.), other candidate characters can be recognized during character recognition. Output with priority over the output. The delimiter of each record is determined by the end bit. In this example,
They are "Shingata (Niigata)" and "My soup".

Reference numeral 122 is a pattern block. By storing a character or a combination of character types to be preferentially output here, the candidate character corresponding to the character recognition is preferentially output. In this example, a combination of "number and number" such as "11:00", a combination of "number and number" such as "2-1", a combination of "first and number" such as "second", and also "Daito number" And the combination of “number and time” such as “1 time”. The record delimiter is determined by the end bit.

FIG. 13 is a diagram showing the structure of a new word dictionary for kana-kanji conversion. This kana-kanji conversion new word dictionary is information used during kana-kanji conversion processing, and stores the readings, headings, etc. of words registered as new words by the user.

In FIG. 13, reference numeral 131 is a reading directory into which the leading 3 characters of the reading of the leading record of each block of the new word dictionary body are assigned. In this case, since the records of the new word dictionary are sorted in ascending order of the reading character codes, a high speed new word dictionary search can be performed by searching the reading directory.

Reference numeral 132 is the main body of the new word dictionary, in which the reading, heading (converted word), and dictionary number are stored.
In this example, the heading "Amaha" is read along with the dictionary number in the reading "Amo". Similarly,
The reading "Waizu" stores the heading "Wise" together with the dictionary number.

FIG. 14 is a diagram showing the structure of a kanji-mixed kana-kanji conversion new word dictionary. This kanji-mixed kana-kanji conversion new word dictionary is information used in the kanji-mixed kana-kanji conversion process, and stores the readings, headings, etc. of the kanji-mixed kanji words registered by the user as new words.

In FIG. 14, reference numeral 141 is a reading directory into which the leading three characters of the reading of the leading record of each block of the new word dictionary body are assigned. Since the records in the new word dictionary are sorted in ascending order of the reading character code, a high speed new word dictionary search can be performed by searching this reading directory.

Reference numeral 142 is the main body of the new word dictionary, in which the reading, heading (converted word), and dictionary number are stored.
In this example, the heading "Amou" is read along with the dictionary number for the reading "Amou". Similarly,
The heading "outer sauce" with the reading "outer sauce" is stored together with the dictionary number.

FIG. 15 is a diagram showing the structure of a character new word dictionary for character recognition. The character new word dictionary for character recognition is a new word dictionary used in character recognition processing, and stores characteristic data of handwritten characters input by the user and a character code corresponding to the handwritten character. In this example,
It shows a case where handwritten characters "i" and "I" are registered as new words.

In FIG. 15, reference numeral 151 is a separator, which divides the record by this information. Code 15
Reference numeral 2 is a character code of "i", which is code data of a character to be registered as a new word. Reference numeral 153 is the handwritten character feature data of "i", which is the feature data of the handwritten character to be registered as a new word.

FIG. 16 is a diagram showing the structure of a character learning dictionary for character recognition. The character new word dictionary for character recognition is a learning dictionary used in character recognition processing, and stores characteristic data of a handwritten character input by a user and a character code corresponding to the handwritten character. In this example,
The case where the handwritten characters of "i" and "I" are learned is shown.

In FIG. 16, reference numeral 161 is a separator, which divides the record by this information. Code 16
Reference numeral 2 is a character code of "i", which is code data of a character to be learned. Reference numeral 163 is handwritten character feature data of "i", which is handwritten character feature data to be learned.

FIG. 17 is a diagram showing the structure of the character string new word dictionary for character recognition. While the new word dictionary described in FIG. 15 targets one character, this character string new word dictionary for character recognition targets a character string. That is, similar characters such as "I" and "1" may be judged only by the preceding and succeeding characters. Therefore, as in this example, by simultaneously storing the character string "Dai 1" and the handwritten feature data of the character "1", it is possible to recognize similar characters "1" and "I".

In FIG. 17, reference numeral 171 is a separator, which divides the record by this information. Code 17
2 is the character code of "dai 1", which is the code of the character to be registered. Reference numeral 173 is handwritten character feature data of "1", which is handwritten character feature data to be registered.

FIG. 18 is a diagram showing the structure of a character string learning dictionary for character recognition. While the learning dictionary described in FIG. 16 targets one character, this character recognition character string learning dictionary targets a character string. That is, similar characters such as "I" and "1" may be judged only by the preceding and succeeding characters. Therefore, as in this example, by simultaneously storing the character string "Dai 1" and the handwritten feature data of "1", it is possible to recognize similar characters "1" and "I".

In FIG. 18, reference numeral 181 is a separator, which divides the record by this information. Code 18
2 is the character code of "dai 1", which is the code of the character to be learned. Reference numeral 183 is handwritten character characteristic data of "1", which is handwritten character characteristic data to be learned.

Next, a method of processing handwritten input characters according to the present invention will be described with reference to specific examples. Here, (1)
Priority determination processing regarding character arrangement, (2) priority determination processing regarding sound reading elements, (3) priority determination processing regarding reverse reading elements, (4) priority determination processing regarding learning, (5) priority determination processing regarding dictionary lookup, (6 ) Priority determination processing regarding grammar, (7) Priority determination processing regarding semantic processing,
(8) Priority determination processing regarding difficult characters and (9) Priority determination processing regarding selection of recognition result will be separately described.

(1) Priority Discrimination Processing Regarding Character Arrangement As shown in FIG. 19A, it is assumed that a recognition candidate as shown in the figure is obtained as a result of character recognition processing of a character string “star”, for example ( Example 1). In such a case, there are very few cases where the alphabet exists before and after, and one hiragana character is sandwiched between them, and conversely, there are many cases where the alphabet is continuous, so "star" is processed first.

Further, as shown in FIG. 19B, it is assumed that a recognition candidate as shown in the figure is obtained as a result of character recognition processing of a character string "12:00", for example (Example 2). In such a case, there are very few cases where numbers exist before and after, and one hiragana character is sandwiched between them, and conversely, there are many cases where numbers, symbols, and alphabets are sandwiched between them. Therefore, "12:00" Is processed first.

(2) Priority Discrimination Processing Regarding On-Speech Reading Element As shown in FIG. 20, for example, as a result of character recognition of a character string "Kanojo search, Yuko ni nerakusaku", "Search, It is assumed that a recognition candidate as shown in the figure is obtained in the "Yuko" part (Example 3). In such a case, it is very rare that the reading element "shu" is separated from "shi, yuu", and the character string "shi, yuu" is 1
Character strings with only characters as second candidates are "ji, yuu", "shi ・
There are four candidates, "Yu" and "Shi, Yutsu", but "Shi Yu"
Except for the same reason (sound reading element), postponed,
"Yu" is processed first.

(3) Priority Discrimination Processing Regarding Reversed-Sound Reading Element As shown in FIG. 21, it is assumed that a recognition candidate as shown in the figure is obtained as a result of character recognition processing of a character string "Kaiju" (Example 4). . In such cases,
Since there are very few cases where "E" is attached after "Ju" and "U" is attached after "JU" in many cases, "KAIJU" is processed first.

(4) Priority Discrimination Processing for Learning As shown in FIG. 22, it is assumed that a recognition candidate as shown in the figure is obtained as a result of character recognition processing of a character string "aisukurimu" (example 5). In such a case, if "Aiscreem" was learned by kana-kanji conversion processing or character recognition processing, it is more likely to be used as "Aiscreem" than "Aizukurim". Is processed first.

(5) Priority Discrimination Processing Regarding Dictionary Lookup As shown in FIG. 23A, it is assumed that a recognition candidate as shown in the figure is obtained as a result of character recognition processing of a character string "Kanji" (example) 6). At this time, "Kashiji"
Not in the kana-kanji conversion dictionary, but "Kanji", "Kashi",
If "Jiji" is in the Kana-Kanji conversion dictionary, it is more likely to be used as "Kanji" than "Kashiji". Therefore, "Kanji" is processed first to speed up the process.

Further, as shown in FIG. 23B, it is assumed that a recognition candidate as shown in the figure is obtained as a result of the character recognition processing of the character string "Tonta" (Example 7). In such a case, if "Tonta" is in the new Kana-Kanji conversion dictionary and "Toyota" is in the Kana-Kanji conversion dictionary and is not in the new word dictionary, "Tonta" is better than "Toyota". Since it has a high probability of being used in, "Tonta" is processed first.

(6) Priority Discrimination Processing Regarding Grammar As shown in FIG. 24, it is assumed that a recognition candidate as shown in the figure is obtained as a result of the character recognition processing of the character string "Dai 1" (Example 8). In such a case, there are a plurality of candidates such as "Dashi 1" in the combination of character recognition candidates, but "Dai" is a numerical prefix and "1" is a number at the final stage of Kana-Kanji conversion. , "Dai 1" is processed first because it has a high probability of being prioritized in grammatical processing.

(7) Priority Discrimination Processing Regarding Semantic Processing As shown in FIG. 25A, it is assumed that a recognition candidate as shown in the figure is obtained as a result of the character recognition processing of the character string "Natsuhaatsu". (Example 9). In such a case, if there is no AI information (including AI learning) for “Noneha Atsushi” and there is AI information for “Natsuha Atsushi”,
Since there is a higher probability that "Natsuha Atsushi" will be used than "Natsuha Atsushi", "Natsuha Atsushi" will be processed first.

Further, as shown in FIG. 25B, it is assumed that a recognition candidate as shown in the figure is obtained as a result of character recognition processing of a character string "Odakyu" (Example 1).
0). In this case, if the contents of the document entered so far are judged to be related to railways, the probability of correct answer is higher in "Odaku" than in "Obaku", so "Odakyu" is processed first. To do.

(8) Priority Discrimination Processing for Difficult Characters As shown in FIG. 26 (a), it is assumed that a recognition candidate as shown in the figure is obtained as a result of character recognition processing of a character string "bonito" (example) 11). Here, "bonito" is difficult to write, so it is often written in hiragana (difficult characters, characters with a large number of strokes, and at the same time, other fish-related characters "bonito").
Or because it's a character that you might memorize by mistake as "餤". Therefore, in such a case, “bonito” is processed before other candidates.

Further, as shown in FIG. 11B, it is assumed that a recognition candidate as shown in the figure is obtained as a result of the character recognition processing of a character string containing kanji characters such as "Shingata" (Example 12). . Here, the "gata" in "Niigata" is hard to write, so it is often written in hiragana. On the other hand, the number of strokes of "shinkata" in "shinkata" is small, so it is often written in kanji. Therefore, in such a case, since there is a higher probability that the correct answer is "shingata" than "shinkata", "shingata" is processed first.

(9) Priority Discrimination Processing Regarding Selection of Recognition Result As shown in FIG. 27A, for example, as a result of character recognition processing of a character string "Ajita", a recognition candidate as shown in the figure is obtained. (Example 13). Here, FIG.
As shown in, the user operates the next candidate key to
Is selected. In such a case, the user's selection is given the highest priority, even if the candidates "Ashita", "Mata", etc., which include "Shi" that have not been selected are high in the priority processing, and those candidates are selected. Drop priority or remove from processing.

Next, a flow chart showing the processing operation of the embodiment will be explained. FIG. 28 is a flow chart showing the processing from the character input to the kana-kanji conversion result output in the document creation device having the kana-kanji conversion function (FIG. 1).

Reference numeral A1 is a step for inputting characters,
Enter handwritten characters with the tablet and pen. Code A
Reference numeral 2 is a step of character recognition processing, and the character recognition processing is continued using the sub CPU even during character input.

Reference numeral A3 is a step of displaying the character recognition result, and displays the character recognition result obtained by the character recognition processing. However, if the processes of the symbols A5 to A12 can be processed at high speed, the process of A7 may be omitted because it is not noticeable to the user.

Reference numerals A4 and A11 are steps of interrupt processing. When the confirmation key is pressed with respect to the character recognition result, an interruption occurs, and the replacement of candidates is stopped for the confirmed recognition result. .

Reference numeral A5 is a step for performing dictionary lookup (reading information) of a kana-kanji dictionary or the like, and it performs dictionary lookup for reading information of the standard dictionary for Kana-kanji conversion processing, various extended dictionaries, various new word dictionaries, and various learning dictionaries.

Reference numeral A6 is a step for performing priority determination processing by reading information, on-speech reading elements, reverse-sound reading elements, priority discrimination dictionary information, various learning dictionaries, various new word dictionaries, and various expansion dictionaries. By this processing, future kana-kanji dictionary Etc., the dictionary lookup processing, the display of character recognition results, the order of kana-kanji conversion processing are determined, and the candidates not to be processed in the future are determined.

The details of the priority determination process performed in A6 and A9 will be described later with reference to the flow charts shown in FIGS. Reference numerals A7 and A10 are steps for redisplaying the character recognition results, and the character recognition results are redisplayed in the order determined by the priority determination processing of the reference numerals A6 and A9.

Reference numeral A8 is a step for performing dictionary lookup (other than reading information) of a kana-kanji dictionary or the like, and information other than the reading information of the standard dictionary for kana-kanji conversion processing, various extended dictionaries, various new word dictionaries, various learning dictionaries (part of speech). , Heading, semantic information, etc.).

Reference numeral A9 is a step for performing priority discrimination processing using information (part-of-speech information, semantic information, etc.) other than dictionary or data reading information relating to kana-kanji conversion, priority discrimination dictionary information, various learning dictionaries, various new word dictionaries, and various extended dictionaries. By this processing, dictionary lookup processing of a kana-kanji dictionary or the like in the future, display of character recognition results, rank of kana-kanji conversion processing, determination of priority candidates for kana-kanji conversion processing and determination of candidates not processed in the future are performed.

Reference numeral A12 is a step for performing kana-kanji conversion processing, which converts the input characters into kana-kanji characters in the order determined by the priority determination processing. Reference numeral A13 is a step for displaying the kana-kanji conversion result, which displays the kana-kanji conversion result.

FIG. 29 is a flow chart showing the processing from the character input to the kana-mixed kana-kanji conversion result output in the document creating apparatus (FIG. 2) having the kanji-mixed kana-kanji conversion function.

Reference numeral B1 is a step for inputting characters,
Enter handwritten characters with the tablet and pen. Code B
Reference numeral 2 is a step of character recognition processing, and the character recognition processing is continued using the sub CPU even during character input.

Reference numeral B3 is a step of displaying the character recognition result, which displays the character recognition result obtained by the character recognition processing. However, when the processes from B5 to B12 can be processed at high speed, the process of B7 may be omitted because it does not catch the eyes of the user.

Reference numerals B4 and B11 are steps for displaying interruption processing. When the confirmation key is pressed for the character recognition result, an interruption occurs and the replacement of candidates for the confirmed recognition result is stopped. To do.

Reference numeral B5 is a step for performing dictionary lookup (reading information) of a kanji / kanji mixed kanji dictionary, etc., and is a dictionary for reading information of kanji / kanji mixed kana / kanji conversion processing standard dictionary, various extended dictionaries, various new word dictionaries, various learning dictionaries. Pull.

Reference numeral B6 is a step for performing priority determination processing by reading information, on-speech reading elements, reverse-sound reading elements, priority discrimination dictionary information, various learning dictionaries, various new word dictionaries, and various extension dictionaries. It determines the order of Kana-Kanji dictionary etc., the display of character recognition results, the order of Kana-Kanji conversion processing, and the candidates not to be processed in the future.

The details of the priority determination process in A6 and A9 will be described later with reference to the flow charts shown in FIGS. Reference characters B7 and B10 are steps for redisplaying the character recognition result, and the character recognition result is redisplayed in the order determined by the priority determination processing of the reference signs B6 and B9.

Reference numeral B8 is a step for performing dictionary lookup (other than reading information) of a kanji / kanji / kanji dictionary or the like, other than the reading information of the kanji / kanji / kanji / kanji conversion processing standard dictionary, various extended dictionaries, various new word dictionaries, various learning dictionaries. The dictionary is searched for information (part of speech, headline, semantic information, etc.).

Reference numeral B9 denotes priority discrimination processing using information (part-of-speech information, semantic information, etc.) other than dictionary or data reading information concerning kanji mixed with kanji, priority discrimination dictionary information, various learning dictionaries, various new word dictionaries, and various extended dictionaries. This is a step, and by this processing, dictionary lookup processing such as Kanji-Kana-Kanji dictionary in the future, display of character recognition results, ranking of Kanji-Kana Kana-Kanji conversion processing, determination of priority candidates for Kanji-Kana Kana-Kanji conversion processing and selection of candidates not to be processed in the future Make a decision.

Reference numeral B12 is a step for performing kana-kanji kana-kanji conversion processing, in which the input characters are kanji-mix kana-kanji conversion in the order determined by the priority determination processing. Code B
Reference numeral 13 is a step of displaying a kanji-mixed kana-kanji conversion result, which displays the kanji-mixed kana-kanji conversion result.

Next, a flow chart showing the priority discrimination processing in the embodiment will be explained. Note that here, description will be made assuming that the priority determination process is performed by the document creation device having the kana-kanji conversion function (steps A6 and A9 in FIG. 28). However, in the document creation device provided with kanji-kana kanji conversion. When priority determination processing is performed (step B in FIG. 29)
The same applies to B6 and B9).

FIG. 30 is a flow chart showing the priority discrimination processing (see FIG. 19) regarding the arrangement of characters. When a plurality of recognition candidates are obtained for each character as a result of the character recognition processing of the character string input by handwriting on a character-by-character basis, first, at step C1, for each character obtained by the character recognition processing. It is checked whether or not there is a candidate according to the character arrangement information stored in the pattern block (12f in FIG. 1, 122 in FIG. 12) of the priority determination dictionary in the combination of recognition candidates.

If there is a corresponding candidate, step C2
In addition to setting the output order of the character recognition result based on the candidate, set the processing order of the kana-kanji conversion, the output order of the kana-kanji conversion result, and determine the candidate to delete to speed up the processing. Try.

FIG. 31 is a flow chart showing the priority discrimination processing (see FIG. 20) regarding the sound reading elements. When a plurality of recognition candidates are obtained for each character as a result of the character recognition processing of the character string input by handwriting on a character-by-character basis, first, at step D1, for each character obtained by the character recognition processing. A delimiter code (space, punctuation mark, etc.) that has a phonetic reading element stored in the phonetic reading element table (12g in FIG. 1 and FIG. 8) in the combination of recognition candidates and that divides a phrase in the middle of the phonetic reading element. )
Check if there is a candidate with.

If there is a corresponding candidate, step D2
In addition to setting the output order of the character recognition result based on the candidate, the processing order of the kana-kanji conversion result, the output order of the kana-kanji conversion result according to the candidate, and determining the candidate to be deleted, Speed up.

FIG. 32 is a flowchart showing the priority discrimination processing (see FIG. 21) regarding the reverse sound reading element. When a plurality of recognition candidates are obtained for each character as a result of the character recognition processing of the character string input by handwriting on a character-by-character basis, first, at step E1, for each character obtained by the character recognition processing. It is checked whether or not there is a candidate having the reverse sound reading element stored in the reverse sound reading element table (12h in FIG. 1, FIG. 9) in the combination of recognition candidates.

If there is a corresponding candidate, step E2
In addition to setting the output order of the character recognition result based on the candidate, the processing order of the kana-kanji conversion result, the output order of the kana-kanji conversion result according to the candidate, and determining the candidate to be deleted, Speed up.

FIG. 33 shows a priority determination process for learning (see FIG. 2).
2)). As a result of character recognition processing for each character string input by handwriting,
When a plurality of recognition candidates are obtained for each character, first, in step F1, various learning dictionaries (13g in FIG. 1, FIG. 3 in FIG. 1) are included in the combination of the recognition candidates for each character obtained by the character recognition processing. ~ 7, Figure 10, Figure 16, Figure 1
It is checked whether there is a candidate that matches the learning content of 8).

If there is a corresponding candidate, step F2
In addition to setting the output order of the character recognition result based on the candidate, the processing order of the kana-kanji conversion result, the output order of the kana-kanji conversion result according to the candidate, and determining the candidate to be deleted, Speed up.

FIG. 34 is a flow chart showing the priority determination process (see FIG. 23) regarding the dictionary lookup. When a plurality of recognition candidates are obtained for each character as a result of the character recognition processing of the character string input by handwriting on a character-by-character basis, first, at step G1, for each character obtained by the character recognition processing. It is checked whether or not there is a candidate in the combination of recognition candidates that matches the registered contents of the Kana-Kanji conversion extended dictionary (12i in FIG. 1) and various new word dictionaries (13h in FIG. 1, FIG. 13, FIG. 15, FIG. 17). .

If there is a corresponding candidate, step G2
In addition to setting the output order of the character recognition result based on the candidate, the processing order of the kana-kanji conversion result, the output order of the kana-kanji conversion result according to the candidate, and determining the candidate to be deleted, Speed up.

FIG. 35 shows a priority discrimination process for grammar (FIG. 2).
4)). As a result of character recognition processing for each character string input by handwriting,
When a plurality of recognition candidates are obtained for each character, first, in step H1, the grammar of the kana-kanji conversion means (12c in FIG. 1) is included in the combination of the recognition candidates for each character obtained by the character recognition processing. Check whether there are candidates that follow the rules.

If there is a corresponding candidate, step H2
In addition to setting the output order of the character recognition result based on the candidate, the processing order of the kana-kanji conversion result, the output order of the kana-kanji conversion result according to the candidate, and determining the candidate to be deleted, Speed up.

FIG. 36 is a flow chart showing the priority discrimination processing (see FIG. 25) regarding the semantic processing. When a plurality of recognition candidates are obtained for each character as a result of character recognition processing of the character string input by handwriting on a character-by-character basis, first, in step I1, for each character obtained by the character recognition processing. Among the combinations of recognition candidates, the rules of the semantic processing of the kana-kanji conversion means (12c in FIG. 1), or the information learned by the automatic co-occurrence learning dictionary (FIG. 6), the neuroactivity value table (FIG. 10), etc. ( Check whether there is a candidate according to 12g) of FIG.

If there is a corresponding candidate, step I2
In addition to setting the output order of the character recognition result based on the candidate, the processing order of the kana-kanji conversion result, the output order of the kana-kanji conversion result according to the candidate, and determining the candidate to be deleted, Speed up.

FIG. 37 is a flow chart showing the priority discrimination process (see FIG. 26) for difficult characters. When a plurality of recognition candidates are obtained for each character as a result of the character recognition processing of the character string input by handwriting on a character-by-character basis, first, at step J1, for each character obtained by the character recognition processing. In the combination of recognition candidates, priority information of characters that are difficult to enter in the priority determination dictionary (12f in FIG. 1,
It is checked whether there is a candidate according to 121) in FIG.

If there is a corresponding candidate, step J2.
In addition to setting the output order of the character recognition result based on the candidate, the processing order of the kana-kanji conversion result, the output order of the kana-kanji conversion result according to the candidate, and determining the candidate to be deleted, Speed up.

FIG. 38 is a flow chart showing the priority discrimination processing (see FIG. 27) regarding the selection of the recognition result. When a plurality of recognition candidates are obtained for each character as a result of the character recognition processing of the character string input by handwriting on a character-by-character basis, first, at step K1, for each character obtained by the character recognition processing. It is checked whether or not the character already defined by the user is present in the combination of recognition candidates.

If there is a corresponding candidate, step K2
In that case, the candidate is given the highest priority, and the output order of the character recognition results is set for the subsequent candidates. In addition, the processing order of the kana-kanji conversion result and the output order of the kana-kanji conversion result are set according to the candidate, and the candidate order to be deleted is set. By making the determination, the processing speed is increased.

Next, the operation of the document creating apparatus in the embodiment will be described. Here, the operations of (a) a document creation device having a kana-kanji conversion function and (b) a document creation device having a kana-kanji kanji conversion function will be described separately.

(A) Document Creation Device Having Kana-Kanji Conversion Function In the document creation device shown in FIG.
5 and the tablet 16 (FIG. 11) are used to manually input the character string to be converted into kana-kanji character by character (FIG. 2).
8 step A1). The handwritten characters input at this time are sequentially stored in the character recognition work buffer 13a and then subjected to character recognition processing by the character recognition means 12a (step A2 in FIG. 28).

The character recognition processing is performed by referring to the character recognition dictionary 12b. That is, by comparing the input character (handwritten character) with the standard character in the character recognition dictionary 12b,
Characters with a high degree of matching are sequentially read from the dictionary 12b, and this is made a recognition candidate for the input character.

Here, the sub CPU 1 for character recognition is used.
By providing 1b, the character can be recognized even during the character input, but the present invention is not limited to the number of CPUs. When a plurality of recognition candidates are obtained for each handwritten character, the character recognition output buffer 13b outputs these recognition candidates according to the output order (output order set for each character unit) set in the process of the recognition process. After being stored in the LCD, it is displayed on the LCD 17 through the input / output control unit 14 (see FIG.
8 step A3).

Here, if a confirmation instruction for selecting a candidate such as pressing the select / execute key 115 is given to the displayed character recognition result, an interrupt occurs (step A4 in FIG. 28). As a result, with respect to the character of which the confirmation is instructed, the replacement of the output order of the recognition candidates by the priority determination means 12e will be stopped in the future (Example 13).

Next, the kana-kanji conversion dictionary 12d for the reading information is looked up by the kana-kanji conversion means 12c (step A5 in FIG. 28). As a result, in addition to the reading information obtained by this dictionary and the priority determination dictionary 12f (FIG. 12), the on-reading element table 12g (FIG. 8), the reverse on-reading element table 12h (FIG. 9), various learning dictionaries 13g (FIG. 3,).
(Fig. 4, Fig. 5, Fig. 6, Fig. 7, Fig. 7, Fig. 16, Fig. 18),
By referring to various new word dictionaries 13h (FIG. 13, FIG. 15, FIG. 17) as necessary, the priority determination means 22e sets the output order of character recognition results, the processing order of kana-kanji conversion, and the output of kana-kanji conversion results. Setting of ranks, determination of candidates to be deleted, etc. are performed (step A6 in FIG. 28, example 1, example 2, example 3, example 4, example 5, example 6, example 7, example 11).

After each recognition candidate whose output order has been changed by the priority discrimination processing is stored in the priority discrimination processing output buffer 13f, the first candidate among them is the LCD 17.
Is displayed again (step A7 in FIG. 28).

Further, the kana-kanji conversion dictionary 12d for information other than reading (part of speech, heading, semantic information, etc.) is searched by the kana-kanji conversion means 12c (step A8 in FIG. 28).

As a result, the information other than the reading (part of speech, headline, semantic information, etc.) obtained from the dictionary and the priority determination dictionary 12
By referring to f (FIG. 12), the priority determining means 22e sets the output order of the character recognition result, sets the processing order of the kana-kanji conversion, sets the output order of the kana-kanji conversion result, and determines the candidate to be deleted. (Step A9 in FIG. 28, Example 1, Example 2, Example 5, Example 8, Example 9, Example 10, Example 11).

The recognition candidates whose output ranks have been changed in the priority determination processing are stored in the priority determination processing output buffer 13f and then displayed again on the LCD 17 (step A10 in FIG. 28).

At this time, if a confirmation instruction for selecting a candidate such as pressing the selection / execution key 115 is given to the displayed character recognition result, an interrupt occurs (step A11 in FIG. 28), and the confirmation instruction is issued. The replacement of the output order of the recognition candidates for the selected character is stopped.

Thereafter, the output buffer 1 for priority determination processing
Kana-Kanji conversion processing is performed by the Kana-Kanji conversion unit 12c in the order of the recognition candidates stored in 3f (step A12 in FIG. 28), and the conversion result is stored in the kana-Kanji conversion processing work buffer 13d, and then input / output control is performed. It is displayed on the LCD 17 through the unit 14 (step A1 in FIG. 28).
3). In this case, first, the character string of the first candidate is preferentially output as the conversion result, and the second candidate character string is output by the second candidate selection operation.
The candidates and others are displayed in order.

As described above, by setting the output order so that the combination of the candidates that are valid as the character string is given priority among the combinations of the recognition candidates for each character obtained by the character recognition processing, as a result, The recognition rate can be improved, and the conversion rate in subsequent Kana-Kanji conversion processing can also be improved,
The overall processing time can be shortened.

(B) Kanji-mixed Kana-Kanji Character Creation Device with Kanji Conversion Function In the document creation device shown in FIG.
Using the tablet No. 5 and the tablet 26 (FIG. 11), a character string to be converted into kanji with kanji is input by handwriting character by character (step B1 in FIG. 29). The handwritten characters input at this time are sequentially stored in the character recognition work buffer 23a and then subjected to character recognition processing by the character recognition means 22a (step B2 in FIG. 29). In this case, the character recognition process is performed by referring to the character recognition dictionary 22b.

The character recognition processing is performed by referring to the character recognition dictionary 22b. That is, by comparing the input character (handwritten character) with the standard character in the character recognition dictionary 22b,
Characters with a high degree of matching are sequentially read from the dictionary 22b, and this is used as a recognition candidate for the input character.

Here, the sub CPU 2 for character recognition is used.
By providing 1b, the character can be recognized even during the character input, but the present invention is not limited to the number of CPUs. Then, when a plurality of recognition candidates are obtained for each handwritten character, the character recognition output buffer 23b outputs these recognition candidates according to the output order (output order set for each character unit) set in the process of the recognition process. After being stored in the LCD 27, it is displayed on the LCD 27 through the input / output control unit 24 (see FIG. 2).
Step B3 of 9).

Here, when the user gives a confirmation instruction such as pressing the select / execute key 115 to the displayed character recognition result, an interrupt occurs (step B4 in FIG. 29). As a result, with respect to the character for which the confirmation is instructed, the replacement of the output order of the recognition candidates by the priority determination unit 22e will be stopped in the future (Example 1).
3).

Next, kana-kanji and kanji conversion means 22c
The dictionary lookup of the Kanji conversion dictionary 22d containing Kanji concerning reading information is performed (step B in FIG. 28).
5). As a result, in addition to the reading information obtained from this dictionary and the priority determination dictionary 22f (FIG. 12), the phonetic reading element table 22
g (FIG. 8), reverse sound reading element table 22h (FIG. 9), various learning dictionaries 23g (FIG. 3, FIG. 4, FIG. 5, FIG. 6, FIG. 7, FIG. 10, FIG. 16, FIG. 18), various new word dictionaries 23h. (Fig. 1
4, FIG. 15, FIG. 17), if necessary,
The priority determination means 22e sets the output order of the character recognition results, the processing order of the kana-kanji conversion processing, the output order of the kana-kanji conversion results, and the determination of candidates to be deleted (FIG. 2).
Step B6 of 9, the example 1, the example 2, the example 3, the example 4, the example 5, the example 6, the example 7, the example 12).

The recognition candidates whose output ranks have been changed in this priority discrimination processing are output to the priority discrimination processing output buffer 23f.
After being stored in, the first candidate among them is displayed again on the LCD 27 (step B8 in FIG. 29).

Also, kana-mixed kana-kanji conversion means 22c
By this, the Kana-Kanji conversion dictionary 22d for information other than reading (part of speech, heading, semantic information, etc.) is looked up (step B8 in FIG. 29).

As a result, the information (part of speech, headline, semantic information, etc.) other than the reading obtained from the dictionary and the priority discrimination dictionary 22 are obtained.
By referring to f (FIG. 12), the priority determining means 22e sets the output order of the character recognition result, sets the processing order of the kana-kanji conversion, sets the output order of the kana-kanji conversion result, and determines the candidate to be deleted. (Step B9 in FIG. 29, Example 1, Example 2, Example 5, Example 8, Example 9, Example 10, Example 12).

The respective recognition candidates whose output ranks have been changed by the priority discrimination processing are stored in the priority discrimination processing output buffer 23f and then displayed again on the LCD 27 (step B10 in FIG. 29).

At this time, if a confirmation instruction for selecting a candidate such as pressing the selection / execution key 115 is given to the displayed character recognition result, an interrupt occurs (step B11 in FIG. 29), and the confirmation instruction is issued. The replacement of the output order of the recognition candidates for the selected character is stopped.

Thereafter, the output buffer 2 for priority determination processing
Kana-kana kana-kanji conversion processing is performed by the kanji-kana kana-kanji conversion means 22c in the order of the recognition candidates stored in 3f (step B12 in FIG. 29), and the conversion result is stored in the kanji-mix kana-kanji kanji conversion processing work buffer 23d. After that, it is displayed on the LCD 27 through the input / output control unit 24 (step B13 in FIG. 29). In this case, first, the character string of the first candidate is preferentially output as the conversion result,
The second and subsequent candidates are sequentially displayed by the user's next candidate selection operation.

As described above, by setting the output order so that the combination of candidates that is valid as a character string is given priority among the combinations of the recognition candidates for each character obtained by the character recognition processing, as a result, The recognition rate can be improved, and the conversion rate in the subsequent Kanji conversion processing that mixes Kanji characters can be improved, and the overall processing time can be shortened.

[0167]

As described above, according to the present invention, the character string information input by handwriting is subjected to character recognition processing in character units,
For all the recognition candidates for each character obtained by the character recognition processing, (1) priority determination processing regarding arrangement of characters, (2) priority determination processing regarding phonetic reading elements, (3) priority determination regarding reverse reading elements Processing, (4) priority determination processing regarding learning, (5) priority determination processing regarding dictionary lookup,
The character recognition result is obtained by performing (6) priority determination processing regarding grammar, (7) priority determination processing regarding meaning processing, (8) priority determination processing regarding difficult characters, or (9) priority determination processing regarding selection of recognition result. Set the output order of
After that, the kana-kanji conversion process or the kana-kanji conversion kana-kanji conversion process was performed according to the output order, so that candidates that are valid as character strings can be preferentially obtained, and as a result, the recognition rate can be improved. The conversion rate in the subsequent kana-kanji conversion processing or kana-kanji mixed kana-kanji conversion processing can be improved, and the overall processing time can be shortened.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a document creation device having a kana-kanji conversion function of the present invention.

FIG. 2 is a block diagram showing a configuration of a document creation device having a kanji / kanji / kanji conversion function of the present invention.

FIG. 3 is a diagram showing a structure of a provisional learning dictionary.

FIG. 4 is a diagram showing a structure of a long-term learning dictionary.

FIG. 5 is a diagram showing the structure of a single-kanji learning dictionary.

FIG. 6 is a diagram showing a structure of an automatic co-occurrence learning dictionary.

FIG. 7 is a diagram showing a structure of a phrase learning dictionary.

FIG. 8 is a diagram showing the structure of a phonetic reading element table.

FIG. 9 is a diagram showing the structure of a reverse sound reading element table.

FIG. 10 is a diagram showing the structure of a neuroactivity value table.

FIG. 11 is a diagram showing a structure of a tablet.

FIG. 12 is a diagram showing a structure of a priority determination dictionary.

FIG. 13 is a diagram showing a structure of a new word dictionary for kana-kanji conversion.

FIG. 14 is a diagram showing the structure of a kanji-mixed kana-kanji conversion new word dictionary.

FIG. 15 is a diagram showing a structure of a character new word dictionary for character recognition.

FIG. 16 is a diagram showing a structure of a character learning dictionary for character recognition.

FIG. 17 is a view showing the structure of a character string new word dictionary for character recognition.

FIG. 18 is a diagram showing the structure of a character string learning dictionary for character recognition.

FIG. 19 is a diagram illustrating a priority determination process regarding a character arrangement.

FIG. 20 is a diagram for explaining a priority determination process regarding a sound reading element.

FIG. 21 is a diagram for explaining a priority determination process regarding a reverse sound reading element.

FIG. 22 is a diagram for explaining a priority determination process regarding learning.

FIG. 23 is a diagram for explaining a priority determination process regarding dictionary lookup.

FIG. 24 is a diagram for explaining priority determination processing regarding grammar.

FIG. 25 is a diagram for explaining priority determination processing regarding semantic processing.

FIG. 26 is a diagram for explaining priority determination processing regarding difficult characters.

FIG. 27 is a diagram for explaining priority determination processing regarding selection of recognition results.

FIG. 28 is a flowchart showing processing from character input to kana-kanji conversion result output in a document creation device having a kana-kanji conversion function.

FIG. 29 is a flowchart showing processing from character input to output of kanji-mixed kana-kanji conversion result in a document creation device having a kanji-mixed kana-kanji conversion function.

FIG. 30 is a flowchart showing a priority determination process regarding the arrangement of characters.

FIG. 31 is a flowchart showing a priority determination process regarding a sound reading element.

FIG. 32 is a flowchart showing a priority determination process regarding a reverse sound reading element.

FIG. 33 is a flowchart showing a priority determination process regarding learning.

FIG. 34 is a flowchart showing a priority determination process regarding dictionary lookup.

FIG. 35 is a flowchart showing priority determination processing regarding grammar.

FIG. 36 is a flowchart showing priority determination processing regarding semantic processing.

FIG. 37 is a flowchart showing priority determination processing regarding difficult characters.

FIG. 38 is a flowchart showing priority determination processing regarding selection of recognition results.

[Explanation of symbols]

11 ... CPU, 11a ... Main CPU, 11b ... Sub C
PU, 12 ... Storage device (1), 12a ... Character recognition means,
12b ... Character recognition dictionary, 12c ... Kana-Kanji conversion means, 1
2d ... Kana-Kanji conversion dictionary, 12e ... Priority determination means, 12
f ... Priority discrimination dictionary, 12g ... Phonetic reading element table, 12
h ... Inverse reading table, 12i ... Kana-Kanji conversion extended dictionary, 13 ... Storage device (2), 13a ... Character recognition work buffer, 13b ... Character recognition output buffer, 13c ... Kana-Kanji conversion work buffer, 13d ... Kana-Kanji conversion output Buffer, 13e ... Work buffer for priority determination processing, 13f ... Output buffer for priority determination processing, 13g ... Various learning dictionaries, 13h ... Various new word dictionaries, 14 ... Input / output control unit, 15 ... Pen, 16 ... Tablet, 17 ... LCD Two
1 ... CPU, 21a ... Main CPU, 21b ... Sub CP
U, 22 ... Storage device (1), 22a ... Character recognition means, 2
2b ... Character recognition dictionary, 22c ... Kanji mixed kana-kanji conversion means, 22d ... Kanji mixed kana-kanji conversion dictionary, 22e ...
Priority discrimination means, 22f ... Priority discrimination dictionary, 22g ... Onyomi reading element table, 22h ... Inverse reading table, 22i ... Kanji mixed kana conversion expansion dictionary, 23 ... Storage device (2), 23a ... Character recognition work buffer, 23b ...
Character recognition output buffer, 23c ... Kana-mixed kana-kanji conversion processing work buffer, 23d ... Kanji-mixed kana-kanji conversion output buffer, 23e ... Priority discrimination processing work buffer, 23f ... Priority discrimination processing output buffer, 23g ...
Various learning dictionaries, 23h ... Various new word dictionaries, 24 ... Input / output control unit, 25 ... Pen, 26 ... Tablet, 27 ... LCD.

Claims (9)

[Claims] 1. A character recognition process of character string information input by handwriting is performed on a character-by-character basis, and then the combination of recognition candidates for each character obtained by this character recognition process Check whether there is a candidate according to the alignment information, and if there is a candidate, set the output order of the character recognition result based on that candidate, and thereafter, according to the output order, kana-kanji conversion processing or kana-kanji mixed kana-kanji A handwritten input character processing method characterized in that conversion processing is performed. 2. A phonetic reading element stored in a phonetic reading element table in a combination of recognition candidates for each character obtained by character recognition processing of character string information input by handwriting on a character-by-character basis. And whether there is a candidate that has a delimiter code that delimits a phrase in the middle of the phonetic reading element, and if there is a corresponding candidate, the output order of the character recognition results is determined based on that candidate. A handwritten input character processing method characterized by setting, and thereafter, performing kana-kanji conversion processing or kana-kanji kana-kanji conversion processing according to the output order. 3. A phonetic reading stored in a reverse reading element table in a combination of recognition candidates for each character obtained by character recognition processing of character string information input by handwriting on a character-by-character basis. Check whether there is a candidate that has elements, and if there is a corresponding candidate, set the output order of the character recognition result based on that candidate, and thereafter, according to the output order, kana-kanji conversion processing or kanji-kanji kana-kanji conversion processing. A handwritten input character processing method, characterized in that 4. The character string information input by handwriting is subjected to character recognition processing on a character-by-character basis, and the combination of recognition candidates for each character obtained by this character recognition processing matches the learning content of various learning dictionaries. Check whether there is a candidate, if there is a candidate, set the output order of character recognition results based on that candidate, and then perform kana-kanji conversion processing or kana-kanji kana-kanji conversion processing according to that output order. A handwritten input character processing method characterized in that 5. After kana-kanji conversion or kana-kanji kana-kanji conversion in a combination of recognition candidates for each character obtained by character recognition processing of character string information input by handwriting on a character-by-character basis. Check whether there is a candidate that matches the registered contents of the extended dictionary and various new word dictionaries, and if there is a corresponding candidate, set the output order of the character recognition result based on that candidate, and then according to the output order. A handwritten input character processing method characterized by performing kana-kanji conversion processing or kana-kanji kana-kanji conversion processing mixed with kanji. 6. A kana-kanji conversion or kana-kanji kana-kanji conversion process in a combination of recognition candidates for each character obtained by character recognition processing of character string information input by handwriting on a character-by-character basis. Check whether there is a candidate according to the grammar rule, and if there is a candidate, set the output order of the character recognition result based on that candidate. After that, according to the output order, kana-kanji conversion processing or kana-kanji mixed kana-kanji A handwritten input character processing method characterized in that conversion processing is performed. 7. A kana-kanji conversion or kana-kanji kana-kanji conversion process in a combination of recognition candidates for each character obtained by character recognition processing of character string information input by handwriting on a character-by-character basis. Check whether there is a candidate according to the rules of semantic processing or learning information of the semantic processing, and if there is a candidate, set the output order of character recognition results based on that candidate. A handwritten input character processing method characterized by performing kana-kanji conversion processing or kana-kanji mixed kana-kanji conversion processing according to the output order. 8. A character recognition process is performed on character string information input by handwriting on a character-by-character basis, and it is difficult to input data stored in a priority determination dictionary in a combination of recognition candidates for each character obtained by this character recognition processing. Check whether there is a candidate according to the priority information of the character, and if there is a candidate, set the output order of the character recognition result based on that candidate, and thereafter, according to the output order, kana-kanji conversion processing or A handwritten input character processing method characterized in that kana-kanji mixed kana conversion processing is performed. 9. After the character recognition information of the character string information input by handwriting is processed character by character, whether or not there is a fixed character in the combination of recognition candidates for each character obtained by this character recognition processing. If there is a corresponding candidate, that candidate is given the highest priority, and the output order of the character recognition results is set for the subsequent candidates, and thereafter kana-kanji conversion processing or kana-kanji kana-kanji conversion processing is performed according to that output order. A handwritten input character processing method characterized by the above.