JPH11282961A - On-line handwritten sentence recognition method, recognition device and storage medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an on-line handwritten sentence recognition method and a recognition device capable of obtaining a correctly recognized result even when rewrite and additional write are performed. SOLUTION: This device is provided with a handwritten data storage part 35 for storing handwritten data as stroke data strings aligned in the order of being inputted from an input device 30, a handwritten data connection part 36 for connecting the generated plural handwritten data and turning them to one stroke data string, and a character recognition part 37 for recognizing a character, based on the one stroke data string connected in the handwritten data connection part 36. The handwritten data connection part 36 attains one stroke data string while maintaining the order among stroke data belonging to the same handwritten data and deciding the order by the positional relationship of strokes with each other among the stroke data belonging to the different handwritten data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus suitable for recognizing on-line characters in a handwritten text, particularly a handwritten text which has been partially rewritten or rewritten many times. And a recording medium.

[0002]

2. Description of the Related Art As an input device for a word processor or an editor, an online handwritten character recognition device using a stylus pen and a coordinate input device is known. The coordinate input device has, for example, an input display unit in which a display device such as a liquid crystal display and a transparent tablet are stacked, and a coordinate point sequence for each stroke that is a minimum unit that constitutes a character from the movement of the stylus pen in contact with the transparent tablet. Get information. The handwritten character input device recognizes the coordinate point sequence information as a character by collating it with a dictionary and converts it into corresponding code data. The handwritten character input device also uses stroke position information and stroke order information as well. It has a feature that the recognition rate is high because it can be performed.

[0003] When a sentence is input by handwriting, it is difficult to input a complete sentence without any mistake, and partial rewriting and rewriting frequently occur. Even for such partially rewritten or rewritten handwritten input sentences, the handwriting input device becomes inconvenient if it cannot be accurately recognized without burdening the operator. Use value is halved. As a conventional method of rewriting a sentence input by handwriting and recognizing the same, Japanese Patent Laid-Open No. 7-140
No. 07, “Online Handwritten Character Recognition Method and Apparatus”
And a method described in Japanese Patent Application Laid-Open No. 7-296103, “Online Character Recognition Apparatus”.

[0004]

SUMMARY OF THE INVENTION However, Japanese Patent Application Laid-Open No. 7-1
The method disclosed in Japanese Patent Application Publication No. 4007 reorders all stroke data in order from the left, and there is a problem that the character recognition rate is reduced because the stroke order data obtained is lost. Also, Japanese Patent Application Laid-Open No. 7-296103
The method of the publication requires that the number of strokes to be rewritten to maintain the stroke order, which loses the natural interface when writing sentences using paper and pen. There's a problem. Also,
Although both can deal with partial rewriting of characters, there is a problem that characters themselves cannot be rewritten.

The present invention has been made in view of such problems of the prior art, and it is possible to freely perform rewriting and rewriting with a natural interface when writing sentences using paper and a pen. It is an object of the present invention to provide an online handwritten sentence recognition method, a recognition device, and a recording medium that are capable of obtaining a correct recognition result from the handwritten data.

[0006]

First, the relationship between handwritten data and handwritten text in the present invention will be described, taking as an example the case where a horizontally written Japanese sentence is input by handwriting. FIG. 1 is a description example of handwritten data. Handwritten data 10
Is stroke data drawn with one stroke (coordinate data that continues from the start position to the end position of one stroke)
And may be composed of a stroke data string that constitutes one character or symbol, or may be composed of a stroke data string that constitutes a plurality of characters or symbols. Also,
The handwritten data may be a stroke data string that forms only a part of one character or symbol. For example, the character "a" in Hiragana consists of a stroke that is drawn horizontally from left to right, a stroke that is drawn vertically from top to bottom, and a curved stroke that resembles the shape of the "no" character. The handwritten data 10 composed of the character string “Ai” shown in FIG. 1 is a data string in which stroke data from stroke to stroke are arranged in order.

FIG. 2 is a description example of a handwritten sentence. The handwritten sentence 20 is composed of one or more handwritten data 10, 11, 12
Is a set of The breaks between the handwritten data 10, 11, and 12 are separated when a certain period of time has elapsed between the input of the last stroke of one piece of handwritten data (for example, the stroke in FIG. 1) and the input of the next stroke. Alternatively, the input may be separated when the stroke input start position is separated from the immediately preceding stroke input position by a certain distance or more. When an operation other than the input such as erasing the stroke data is performed, the handwritten data is separated there. The present invention relates to handwritten data 1
A means for judging a joint position between 0,11,12, し て, and joining handwritten data 10,11,12, ‥‥ at the position, and a means for recognizing the combined handwritten data 20 with characters; It is characterized by being.

That is, according to the present invention, in an online handwritten text recognition method for recognizing a text input by handwriting, a first step of creating handwritten data as a stroke data sequence arranged in the order of input, and And a second step of combining a plurality of handwritten data created in the step into one stroke data string.

The second step is to maintain the order between stroke data belonging to the same handwritten data,
In addition, the order of stroke data belonging to different handwritten data is determined based on the mutual positional relationship of the strokes to form one stroke data string. When the handwriting input is performed horizontally from left to right, the second step is to maintain the order between the stroke data belonging to the same handwriting data and to maintain the order between the stroke data belonging to different handwriting data. It is preferable to determine the order based on the positional relationship of the left end of the strokes and to make one stroke data string.

In the first step, a new stroke is input when a new stroke is input at a position more than a predetermined distance from the position of the last input stroke, or after an operation other than input of stroke data. When input, new handwritten data starting from the new stroke is created. As described above, the method of dividing the handwritten data may be used together with the method of dividing the handwritten data when a predetermined time elapses from the last stroke to the next stroke input. Thus, character recognition is performed based on one stroke data string created by combining a plurality of handwritten data.

According to the present invention, there is provided an online handwritten sentence recognition apparatus for recognizing a handwritten input sentence, wherein an input unit for inputting the handwritten sentence, and a handwritten data as a stroke data sequence arranged in the order of input from the input unit. , A handwritten data combining unit that combines a plurality of handwritten data created by the handwritten data creating unit into one stroke data string, and one stroke data combined by the handwritten data combining unit A character recognition unit that recognizes a character based on the sequence. The handwritten data combining unit determines the order between stroke data belonging to the same handwritten data while maintaining the order between stroke data belonging to different handwritten data, and determines the order based on the mutual positional relationship of the strokes. It can be.

When the handwriting input is performed in horizontal writing from left to right, the handwriting data combining unit maintains the order between the stroke data belonging to the same handwriting data, and the stroke data belonging to different handwriting data. It is preferable to determine the order between the data based on the positional relationship between the left ends of the strokes to form one stroke data string. The handwritten data creation unit is configured such that when a new stroke is input at a position at least a predetermined distance from the position of the last input stroke, or when a new stroke is input after an operation other than input of stroke data. At times, it may have a function of creating new handwritten data starting from the new stroke.

The above-described online handwritten sentence recognition apparatus according to the present invention performs character recognition by reconstructing handwritten data when partial rewriting or rewriting is performed on a handwritten sentence input from an input unit. It has a function. Further, according to the present invention, there is provided a first method for creating handwritten data as a stroke data sequence arranged in the order of input.
And the plurality of handwritten data created in the first step, while maintaining the order between the stroke data belonging to the same handwritten data, and the position of the mutual stroke between the stroke data belonging to different handwritten data. And a second step of determining an order based on a relationship and combining the strokes into one stroke data string. The computer-readable recording medium stores a program for causing a computer to execute the online handwritten sentence recognition method.

According to the present invention, discontinuous handwritten data can be combined without losing information on the stroke order inputted by handwriting. Even if only the text is rewritten and the text is continuously input, the data input by handwriting can be correctly recognized. Also,
Even when a new character is added between characters, character recognition can be performed correctly.

[0015]

Embodiments of the present invention will be described below with reference to the drawings. Here, an example in which a Japanese sentence is input by handwriting in horizontal writing will be mainly described. FIG.
1 is a block diagram of an example of an online handwritten sentence recognition system to which the present invention has been applied. This online handwritten text recognition system includes an input device 30 for inputting handwritten text, a character recognition processing device 31 for character recognition of handwritten text, a display processing device 32 for displaying a character recognition result, and a character recognition device. From the dictionary 39. The character recognition processing device 31 includes a handwritten data creation unit 33 that creates handwritten data as a stroke sequence arranged in the order input from the input device 30, and a handwritten sentence storage unit that stores handwritten data created by the handwritten data creation unit 33. 35, a handwritten data combining unit 36 that determines a joining position of the handwritten data included in the handwritten sentence stored in the handwritten sentence storage unit 35 and combines them, and a handwritten sentence combined by the handwritten data combining unit 36. Character recognition unit 3 for character recognition
7 and a character recognition result display section 38 for displaying the result of character recognition by the character recognition section 37 on the display device 32.

The input device 30 has an input display section in which a display device such as a liquid crystal display and a transparent tablet are laminated, and is capable of inputting characters and figures by tracing the input surface of the tablet with a stylus pen, or a liquid crystal projector. Any device can be used as long as it can output handwritten input characters and graphics as coordinate data, such as those that input by tracing the surface of a large coordinate input device such as an electronic board on which an image is projected with a pen. be able to. The characters and figures input by handwriting are displayed once in the handwritten state on the input display unit, and the operator can confirm his / her own input state. Input device 30
In addition, by selecting an icon such as an eraser displayed on the input display unit with a pen and using the function of the eraser, by tracing over a part of a previously input handwritten character string or character, You can erase entire characters or parts of characters.

As shown schematically in FIG. 4, the handwritten text storage unit 35 includes a text storage unit 41 and a handwritten data storage unit 4.
2. A stroke data storage unit 43 is provided. The handwritten data creation unit 33 calculates the information of the individual strokes SK1, SK2,..., SKn input from the handwritten sentence input device 30, that is, the coordinates of the position where the pen is pressed, and the trajectory of the movement while the pen is pressed. The stroke data storage unit 43 analyzes the coordinate sequence information to be displayed and the information on the coordinates of the position where the pen is released, and stores the stroke order, that is, the drawn order. The handwritten data creation unit 33 also includes a handwritten data storage unit 42.
The information about which strokes are formed from one stroke to the next, that is, the information about the break of the handwritten data is stored in the first handwritten data. For example,
Handwritten data SD1 is composed of strokes SK1 to SK7, and handwritten data SD2 is composed of strokes SK8 to SK9,
When the handwritten data SD3 is composed of the strokes SK10 to SK15, a pointer to the strokes SK1 to SK7 is stored at the location of the handwritten data SD1, and a pointer to the strokes SK8 to SK9 is stored at the location of the handwritten data SD2. Stroke S in place of SD3
A pointer to K10 to SK15 is stored.

In the text storage section 41, the handwritten texts ST1, ST2,
情報, STn is stored with information indicating what handwritten data is constituted. For example, the handwritten text ST1 is composed of the handwritten data SD1, SD2, SD3, and the handwritten text ST2 is composed of the handwritten data SD4, SD5.
In the case of being constituted by SD6 and SD7, handwritten data SD1 to SD3 are placed at the place of the handwritten text ST1.
And the pointers to the handwritten data SD4 to SD7 are stored at the location of the handwritten text ST2. By being able to store a plurality of handwritten sentences ST1, ST2,..., STn, it is possible to handle handwritten sentences composed of a plurality of lines and vertically written sentences. The break of the sentence is determined by displaying the ruled line on the display device 30 and dividing the line by the position of the ruled line, or by dividing the line by the distance between the position of the rightmost stroke of the previous line and the position of the leftmost stroke of the next line. be able to.

When the entire character or a part of the character is traced by using the eraser function provided in the input device 30, the handwritten data creation unit 33 stores the corresponding stroke information in the stroke data. Part 4
Erase from 3. Although the stroke data storage unit 43 and the handwritten data storage unit 42 are separately provided here, the strokes SK1, SK2,.
直接 and デ ー タ data are directly stored in the stroke data storage unit 4.
3 may be omitted.

FIG. 5 shows an example in which a sentence "Today is full moon but dark" is input by hand. FIG. 5A shows a sentence to be input, and FIG.
(E) illustrates the state of input over time. The state of the stroke data storage unit 43 of the handwritten sentence storage unit 35 will be described with reference to FIG. FIG. 6 schematically shows the stroke data in the form of the stroke itself, but actually stores the coordinate data of the stroke.

In the example shown here, as shown in FIG. 5 (b), when the character is input to the middle of the character "dark", there is an error in the character corresponding to "man" which has already been written. Noticed. Therefore, the text input operation was temporarily stopped. FIG.
(A) schematically shows the state of the stroke data storage unit 43 at this time, and data of each stroke is stored in the order of input from the stroke SK1 to the stroke SK41. Next, as shown in FIG. 5C, the left portion of the erroneous character was erased by using the eraser function displayed on the input display section of the input device. As a result of this operation, the stroke data storage unit 43 clears the erased “Soil” stroke data SK12 to SK14 as shown in FIG.
Stroke data of 15 or less is sequentially carried up by three strokes.

Subsequently, as shown in FIG. 5 (d), "sanzui" was written in the erased "earth", and the erroneous part was rewritten. At this time, the stroke data storage unit 43
Then, as shown in FIG. 6 (c), the input stroke data of "sanzui" is added after the stroke SK38. Further, as shown in FIG. 5E, the continuation of the text was input by handwriting. At this time, in the stroke data storage unit 43, as shown in FIG. 6D, the stroke data following the sentence is added after the data of the strokes SK39 to SK41 of "sanzui".

FIG. 7 illustrates the cuts in the handwritten data of the handwritten text input in the order shown in FIG. If the handwritten data is classified using the distance from the position where the immediately preceding stroke was input, the handwritten sentence in this case is divided into a portion 71 from the first input and erasing, and a rewritten portion 72. And the added part 73
Of three handwritten data. These three handwritten data 71, 72, 73 correspond to the strokes SK1 to SK38 and the strokes SK39 to SK4 shown in FIG.
1, corresponding to the strokes SK41 to SK57.

FIG. 8 is a flowchart for explaining the function of the handwritten data combining unit 36. Handwritten sentence storage unit 35
, One handwritten data is extracted from the handwritten text ST1 stored in the storage device, and is used as data D1 (S81). Then, it is determined whether or not the next handwritten data exists (S
82) If the next handwritten data exists, one next handwritten data is taken out and set as data D2 (S2).
83). Next, the result of performing the combining process of the data D1 and the data D2 as described later is stored in the data D1 (S
84), returning to step 82. If it is determined in step 82 that the next handwritten data does not exist, it is determined that all the handwritten data has been combined, and the process of the handwritten data combining unit 36 is terminated. At this time, data D
The handwritten data stored in No. 1 is the combination result.

The case where the handwritten text shown in FIG. 7 is processed according to the flowchart shown in FIG. 8 will be described as follows. In step 81, the handwritten data including the strokes SK1 to SK38 is extracted and set as data D1. In the determination in the next step 82, since there is still handwritten data, the flow proceeds to step 83, where SK3
The next handwritten data consisting of 9 to SK41 is extracted and set as data D2. Thereafter, in step 84, the data D1
And data D2, and the combined data is referred to as D1. In the determination returned to step 82, the process proceeds to step 83 because handwritten data still exists, and the stroke SK4
The next handwritten data consisting of 2 to SK57 is taken out, and is set as data D2. Thereafter, in step 84, the data D1 and the data D2 are combined, and the combined data is referred to as D1.
And In the determination returning to step 82 again, the process ends because there is no next handwritten data.

FIG. 9 is a flowchart for explaining in more detail the combining process of step 84 shown in FIG. Here, an example in which the input is performed horizontally from left to right will be described. First, the handwritten data D3 is initialized in order to secure an area for storing the combined result (S3).
91). Next, the X coordinate of the left end of the handwritten data D1 is compared with the X coordinate of the left end of the handwritten data D2 (S92). If the X coordinate of the left end of the handwritten data D1 is on the left side, the data D1 is set in the data D4. And the data D5
2 is set (S93). In step 92, data D2
If the left end is determined to be on the left side, the data D2 is set in the data area D4, and the data D1 is set in the data area D5 (S94).

Next, the first stroke is extracted from the data D4, and is set as a stroke S (S95). Next, it is determined whether or not the next stroke exists in the data D4 (S96). If the next stroke exists, the X coordinate of the left end of the stroke S is compared with the X coordinate of the left end of the data D5 (S9).
7) If the left end of the stroke S is on the left, the data D
3 is added to the stroke S (S98), the next stroke is extracted from the data D4, set as the stroke S (S99), and the process returns to step 96.

If it is determined in step 97 that the left end of the data D5 is on the left, all the strokes of the data D5 are added to the data D3 (S100).
3 and all the strokes after S of data D4 are added (S
101) and ends. In step 96, the data D
If it is determined that the next stroke does not exist in No. 4, the stroke S is added to the data D3 (S102), and then all the strokes of the data D5 are added to the data D3 (S103), and the processing is ended. At this time, the content stored in the data D3 is the combination result.

The stroke arrays SK1 to SK1 shown in FIG.
SK57, that is, the handwritten data 71, 72, 7 in FIG.
8 and FIG. 9, the handwritten data 72 is inserted in the middle of the handwritten data 71, and the handwritten data 73 is added after the handwritten data 71, as shown in FIG. Reconstructed into data. The reconstructed handwritten data is sent to the character recognition unit 37.
Sent to The character recognition unit 37 cuts out the sent stroke data string in character units, and recognizes the stroke data in each character unit as a character by performing matching with the dictionary 39. The recognized character string is displayed on the display device 32 via the character recognition result display unit 38.

FIG. 10, FIG. 11, and FIG. 12 are explanatory diagrams of another example in which a correction is made during input of a handwritten sentence. Each square in the figure represents a character string composed of one character or a plurality of characters.

FIG. 10 is an explanatory diagram showing an example in which a part of an input character or character string is deleted and corrected to another character or character string. First, as shown in FIG. 10A, characters or character strings are represented by A1, A2, A3,
As a result of handwriting input of A4, the user noticed that there was an input error in A3, erased A3 as shown in FIG. 10B, and subsequently entered a character or a character at the position of A3 as shown in FIG. Column B was re-entered. In this case, the final handwritten input sentence is composed of handwritten data 100a and handwritten data 100b, as shown in FIG. Therefore, according to the present invention, the two handwritten data 100a, 100a
FIG. 10B that is intended by the operator.
The handwritten input sentence can be recognized in the state (c) (the state as displayed on the input display unit).

FIG. 11 shows a case where a plurality of portions A2 and A4 of a character or character string A1 to A5 input first are erased, and those portions are replaced with a character or character string B1, respectively.
It is explanatory drawing which shows the example at the time of rewriting to B2. In this case, the final handwritten input sentence is composed of three handwritten data 110a, 110b, and 11 as shown in FIG.
0c, the handwritten data 110a,
By combining 110b and 110c, a handwritten input sentence can be recognized as if a handwritten sentence was input from the beginning as shown in FIG. 11 (c).

FIG. 12 is an explanatory diagram showing an example in which another character or character string B is added between input characters or character strings A1 and A2. In this case, the final handwritten input sentence is composed of two handwritten data 120a and 120b as shown in FIG. 12 (b), and by combining the handwritten data 120a and 120b according to the present invention, from the beginning, FIG. The handwritten input sentence can be recognized as if the handwritten sentence was input in the state shown in (b).

Although the case where a horizontally written Japanese sentence is input by hand has been described above, the present invention is also applicable to a vertically written Japanese sentence. When combining the vertically written handwritten data, the Y coordinate of the upper end may be compared instead of the X coordinate of the left end. The present invention is also applicable to input of sentences in languages other than Japanese such as English. 13 and 14 are explanatory diagrams showing examples in which the present invention is applied to English input. FIGS. 15 and 16 are explanatory diagrams showing examples in which the present invention is applied to the vertical writing of Japanese sentences.

In the example of FIG. 13, when the user first inputs as shown in FIG. 13 (a), he notices that "a" in the text is an error of "an". Like "a"
"N" is added to "an".
In this case, the final handwritten sentence is composed of two handwritten data 130a and 130b as shown in FIG.
Can be recognized from the beginning as if the article "an" had been input by hand.

The example of FIG. 14 shows a case where the word “an” in the input text has been rewritten as another word “my” as shown in FIG. In rewriting, first, as shown in FIG. 14B, the word "an" is erased, and the word "my" is written in the empty place as shown in FIG. 14C. In this case, the final handwritten sentence is shown in FIG.
As shown in the figure, two handwritten data 140a, 140b
The handwritten data 140a, 14
By combining 0b, the handwritten input sentence can be recognized as if the handwritten sentence was input from the beginning as shown in FIG. 14 (c).

FIG. 15 is an explanatory diagram showing an example in which the present invention is applied to input of a vertically written Japanese sentence. First, FIG.
As shown in (a), "character recognition" was input in vertical writing.
At this time, in the sentence storage unit 41 of the handwritten sentence storage unit 35, a pointer to handwritten data of a sentence consisting of one character “sentence” is stored in ST1, and similarly, ST2, ST
In ST3 and ST4, pointers to handwritten data of a sentence consisting of one character of "character", "recognition", and "knowledge" are stored.
Each handwritten data is composed of a stroke data string as described above. Next, as shown in FIG.
It is assumed that the character "" is added by handwriting between "" and "". The data of the added "" is stored in ST5 of the text storage unit 41.

Thereafter, the handwritten data is reconstructed according to the present invention. In reconstructing the handwritten data, the units of the handwritten sentences ST1, ST2,.
The order of each handwritten sentence is determined by comparing the upper end Y coordinate of each handwritten data in ST3 and #. As a result, the sentence storage unit 41 finally stores “sentence” as shown in FIG.
→ ST1, “character” → ST2, “no” → ST3, “recognition” →
As shown in ST4, "sense" → ST5, the character "" is inserted and stored at the position intended by the operator. Character recognition is performed on the reconstructed handwritten text, and the result is displayed on the display device 32.

FIG. 16 is an explanatory diagram showing an example in which a Japanese sentence input in a vertical writing mode is rewritten. First, as shown in FIG.
A sentence consisting of characters was entered vertically. At this time, in the sentence storage unit 41 of the handwritten sentence storage unit 35, a pointer to the handwritten data of a sentence consisting of one character "sentence" is stored in ST1. A pointer to the handwritten data of a sentence consisting of one character of "test" and "knowledge" is stored. Next, FIG.
As shown in FIG. 16B, it is assumed that the portion on the right side of the character "" is erased and the portion is rewritten as shown in FIG. 16C. At this time, the handwritten data pointed to by the pointer in ST3 holds only the data of the “word” that has not been erased in the character of “test”, and the data of the newly added “shinobi” is ST5. Is stored in

In this state, the handwritten data is reconstructed according to the present invention. In reconstructing the handwritten data, the units of the handwritten sentences ST1, ST2, ST3,.
Upper end Y of each handwritten data of T1, ST2, ST3, ‥‥
The order of each handwritten sentence is determined by comparing the coordinates. As a result, the handwritten data of ST3 and ST5 in the text storage unit 41 shown in FIG. When combining the handwritten data, the left end X coordinate of each stroke is compared, and the "word" part and the "shinobi" part are combined.
As shown in FIG. 6 (d), the character "A" as a result of the combination is stored in ST3. Character recognition is performed on the reconstructed handwritten text, and the result is displayed on the display device 32.

[0041]

According to the present invention, when a sentence is input by handwriting, even if it is erroneously described, only the erroneous part can be rewritten when the error is noticed, and the sentence can be continued. It is possible to correctly recognize a handwritten sentence that has been input, just as if writing with a paper, pen and eraser.

[Brief description of the drawings]

FIG. 1 is a diagram showing an example of handwritten data.

FIG. 2 is a diagram showing an example of a handwritten sentence.

FIG. 3 is a block diagram of one embodiment of a handwritten sentence recognition system to which the present invention is applied.

FIG. 4 is a schematic diagram of a handwritten sentence storage unit.

FIG. 5 is a diagram illustrating an example of a case where a sentence is input by handwriting.

FIG. 6 is a diagram illustrating a state of a stroke data storage unit of the handwritten sentence storage unit.

FIG. 7 is an explanatory diagram showing a break in handwritten data of a handwritten sentence.

FIG. 8 is a flowchart showing a procedure for processing a handwritten sentence.

FIG. 9 is a flowchart showing the contents of a combining process.

FIG. 10 is an explanatory diagram of another example in which a correction is made during input of a handwritten sentence.

FIG. 11 is an explanatory diagram of another example in which a correction is made during input of a handwritten sentence.

FIG. 12 is an explanatory diagram of another example in which a correction is made during input of a handwritten sentence.

FIG. 13 is an explanatory diagram showing an example in which the present invention is applied to English input.

FIG. 14 is an explanatory diagram showing an example in which the present invention is applied to English input.

FIG. 15 is an explanatory diagram showing an example in which the present invention is applied to input of a vertically written Japanese sentence.

FIG. 16 is an explanatory diagram showing an example of a case where a Japanese sentence input in vertical writing is rewritten.

[Description of Signs] 10, 11, 12: Handwritten data, 20: Handwritten text,
Reference numeral 30: handwritten data input device, 31: character recognition processing device, 32: display device, 33: handwritten data creation unit, 35
... handwritten text storage unit, 36 ... handwritten data combining unit, 37
... character recognition unit, 38 ... character recognition result display unit, 39 ... dictionary, 41 ... sentence storage unit, 42 ... handwritten data storage unit, 4
3. Stroke data storage unit, 71, 72, 73 ... handwritten data, 100a, 100b, 110a, 110b,
110c, 120a, 120b, 130a, 130b,
140a, 140b ... handwritten data

 ──────────────────────────────────────────────────の Continuing on the front page (72) Inventor Yasuhiro Nakada 6-81 Onoecho, Naka-ku, Yokohama-shi, Kanagawa Prefecture Hitachi Software Engineering Co., Ltd.

Claims (11)

[Claims] 1. An on-line handwritten sentence recognition method for recognizing a sentence input by handwriting, comprising: a first step of creating handwritten data as a stroke data sequence arranged in the order of input; A second step of combining a plurality of pieces of handwritten data obtained into one stroke data string. 2. The online handwritten sentence recognition method according to claim 1, wherein said second step maintains the order between stroke data belonging to the same handwritten data, and further comprises maintaining the order between stroke data belonging to different handwritten data. In the online handwritten sentence recognition method, the order is determined according to the positional relationship between the strokes to form one stroke data string. 3. The online handwritten sentence recognition method according to claim 1, wherein said second step maintains the order between stroke data belonging to the same handwritten data, and further comprises maintaining the order between stroke data belonging to different handwritten data. An online handwritten sentence recognition method characterized in that an order is determined based on the positional relationship between the left ends of the strokes to form one stroke data string. 4. The on-line handwritten sentence recognition method according to claim 1, wherein the first step includes newly setting a position at least a predetermined distance from the position of the last input stroke. When a new stroke is input or when a new stroke is input after an operation other than input of stroke data, new handwritten data starting from the new stroke is created. . 5. The online handwritten sentence recognition method according to claim 1, wherein character recognition is performed based on one stroke data string created by combining the plurality of handwritten data. An online handwritten sentence recognition method comprising: 6. An on-line handwritten sentence recognition apparatus for recognizing a handwritten input sentence, wherein an input unit for inputting the handwritten sentence, and handwritten data is created as a stroke data sequence arranged in the order of input from the input unit. A handwritten data creating unit, a handwritten data combining unit that combines a plurality of handwritten data created by the handwritten data creating unit into one stroke data string, and a single stroke data string combined by the handwritten data combining unit An online handwritten sentence recognition device, comprising: a character recognition unit that recognizes a character based on a sentence. 7. The on-line handwritten sentence recognition apparatus according to claim 6, wherein said handwritten data combining unit maintains the order between stroke data belonging to the same handwritten data, and between stroke data belonging to different handwritten data. An online handwritten sentence recognition apparatus characterized in that the order is determined based on the mutual positional relationship of strokes to form one stroke data string. 8. The online handwritten sentence recognition apparatus according to claim 6, wherein the handwritten data combining unit maintains the order between stroke data belonging to the same handwritten data, and maintains the order between stroke data belonging to different handwritten data. An online handwritten sentence recognition apparatus characterized in that an order is determined based on the positional relationship of the left ends of the strokes to form one stroke data string. 9. The online handwritten sentence recognition apparatus according to claim 6, wherein the handwritten data creation unit newly sets the handwritten data at a position at least a predetermined distance from the position of the last input stroke. When a new stroke is input, or when a new stroke is input after an operation other than input of stroke data, new handwritten data starting from the new stroke is created. . 10. The on-line handwritten sentence recognition device according to claim 6, wherein said handwritten sentence input from said input unit is partially rewritten or rewritten. An online handwritten sentence recognition apparatus having a function of reconstructing handwritten data and performing character recognition. 11. A first step in which handwritten data is created as a stroke data string arranged in the order of input, and a plurality of handwritten data created in the first step are transferred between stroke data belonging to the same handwritten data. A second step of determining the order of stroke data belonging to different pieces of handwritten data based on the positional relationship between the strokes and combining them to form one stroke data string. A computer-readable recording medium on which a program for causing a computer to execute a sentence recognition method is recorded.