JPH0757053A - Method for displaying handwriting input - Google Patents

Abstract

PURPOSE:To surely obtain a desired recognition result by instructing a correction unless the intended recognition result is obtained and generating handwriting having segmentation matching the correction instruction. CONSTITUTION:If a character 'ten' shown in a figure is inputted although a character 'dai' should be inputted, it is decided that the input instructs the correction of character segmentation and if it is decided that the recognition result matching the correction instruction is not obtained, the handwriting having the character segmentation matching the correction instruction is generated. Namely, a CPU sections the character segmentation again at this time corresponding to the correction instruction, reads handwriting corresponding to the corrected character segmentation out of a handwriting data base, and decides it as handwriting included in this character segmentation. Through the character recognition of this, a recognition result as the character 'dai' is obtained.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention relates to an electronic notebook device,
The present invention relates to a handwriting input display method suitable for use in a portable computer or the like.

[0002]

2. Description of the Related Art Recently, a device such as an electronic notebook device which can input characters by handwriting is becoming popular. Unlike a normal word processor, a device that can input characters by handwriting can input characters by writing them on the board using a pen instead of the keyboard. You can easily enter characters as if you were writing them with a pencil. Compared with the case of using a keyboard, it is not necessary to store the key positions and the like, so that even a beginner can easily input characters.

As described above, in a device capable of handwriting a character, a device having an input frame for inputting a character and a device having no input frame can input a character at an arbitrary position. There is something.

In an apparatus provided with an input frame, the user manually inputs characters one by one in the input frame. This input frame is usually divided into a plurality of areas, and the user inputs one character in each area. Since the input position is limited, the operability will be poor for the user by that amount, but since the area for inputting each character is divided, it is necessary to cut out the input characters one by one. Instead, it is possible to recognize each character relatively accurately.

On the other hand, in a device that does not have an input frame, the user inputs characters by handwriting at an arbitrary position on the input board (tablet). In this case, the character input by handwriting was recognized as a character at a predetermined timing.

That is, when a character is input, the boundary between the input character and the character is determined. Then, each cut out character is recognized as one character, and the recognition result is displayed.

As a result, for example, as shown in FIG. 39 (a), when the user inputs the kanji "Akira", the two characters "day" and "month" constitute one character. In the case of being cut out as a character to be processed, the input character is recognized as the Chinese character of "my".

However, as shown in FIG. 39 (b), the characters "day" and "month" are relatively widened, so that these characters are cut out as separate characters. In such cases, "day" and "month"
Will be recognized as separate characters.

[0009]

In the conventional device, the result recognized as the characters of "bright" or the result recognized as the characters of "day" and "month" is the user's intention. If it is different, you have to enter it again,
There was a problem that it took time to input the intended characters.

The present invention has been made in view of such a situation, and it is possible to more quickly input a desired character, figure, or the like.

[0011]

A handwriting input display method of the present invention is to input handwriting by handwriting, store the input handwriting, divide the input handwriting into segmentation,
It recognizes the segmentation, displays the input handwriting and the recognition result obtained as a result of recognizing the segmentation.When the recognition result is different from the predetermined one, the segmentation is changed, and when the segmentation is changed, it is stored. It is characterized in that a predetermined handwriting is read from existing handwritings, a handwriting corresponding to the changed segmentation is newly generated, and the changed segmentation is recognized.

The change of the segmentation can be performed by a gesture. In addition, changing segmentation can be performed by enlarging or reducing the input handwriting,
It can either be deleted, moved, or transformed.

[0013]

In the handwriting input display method of the present invention, the segmentation is changed when the display of the recognition result obtained by recognizing the segmentation is different from the intended display. Then, at this time, a handwriting corresponding to the changed segmentation is newly generated. And
The segmentation containing the newly generated handwriting is recognized again. Therefore, it is possible to reliably obtain the intended recognition result without re-inputting the handwriting input by handwriting.

[0014]

1 is a block diagram showing the configuration of an embodiment of an apparatus to which the handwriting input display method of the present invention is applied. C
The PU 1 controls each unit according to a program stored in the ROM 2 and executes a character input display operation. The RAM 3 stores predetermined data and the like as needed. The backup RAM 4 is always supplied with a predetermined amount of power from the battery 5, and stores the information to be stored even when the power of the apparatus is turned off. The timer 6 is a crystal oscillator 7
With reference to, the battery 5 is always operated.

The display controller 11 expands the data input via the bus line 8 into the V-RAM 12, and displays the expanded image on the display unit 13. The input tablet 15 is
By operating the pen 16, the input coordinate position data is output to the tablet interface 14, and the tablet interface 14 outputs this data to the bus line 8.
Via the CPU 1 or the display controller 1
It is designed to output to 1.

The microphone 18 outputs the collected audio signal to the solid-state recording processing circuit 17 and stores it. The audio signal output by the solid-state recording processing circuit 17 is supplied to the speaker 21 via the adder 20 and emitted. The dialer 19 is a CPU
It is controlled to 1 and outputs a predetermined dial signal to the speaker 21 via the adder 20.

The plurality of key switch groups 22 are operated when switching the operation of the apparatus, and a signal corresponding to the operation is sent from the key interface 23 to the CP via the bus line 8.
It is designed to be output to U1. Further, by operating a predetermined one of the plurality of key switch groups 24, the power supply circuit 25 is turned on to supply electric power to each unit, and recording, reproduction, stop by the solid-state recording processing circuit 17,
It is designed to be able to perform volume adjustment and so on.

Further, the socket 26 has an IC (not shown).
It is designed so that a card can be inserted.
The C card interface 27 is designed to send and receive data to and from this IC card.

A modem, for example, can be connected to the pin connector 29, and signals can be exchanged with the extension interface 28 via this modem.

The input tablet 15 is made of a transparent material and is formed on the display unit 13. As a result, the display on the display unit 13 can be seen through the input tablet 15.

Next, the operation of the embodiment shown in FIG. 1 will be described with reference to FIG.
This will be described with reference to the flowchart in FIG. When a predetermined character is input by handwriting on the input tablet 15 with the pen 16, the handwriting data is supplied to the RAM 3 via the tablet interface 14 and stored therein. This data is read out by the display controller 11 and the V-RA
It is supplied to the M12 and developed as bit map data. Further, the data stored in the V-RAM 12 is read by the display controller 11, supplied to the display unit 13, and displayed. In this way, the handwriting of the pen 16 is displayed on the display unit 13.

It should be noted that this input is performed at an arbitrary position on the input tablet 15, and is performed in a state where a so-called input frame is not provided. Then, the input handwriting data is stored in the handwriting handwriting database DB1 of the RAM3.

When the input is made in this way, the CPU 1
First, in step S1, character recognition processing is executed. At this time, the CPU 1 reads the handwriting from the handwriting handwriting database DB1 and uses this handwriting as the character recognition 1
It is divided into character segments as units. Then, character recognition is performed for each character segment, and the recognition result obtained as a result of the recognition is supplied to and stored in the recognition result database DB2 of the RAM3.

Next, in step S2, of the recognition results obtained as described above, the first recognition result is read from the recognition result database DB2 and displayed on the display unit 13. That is, the CPU 1 calculates the similarity (score) with respect to a predetermined character stored in advance for each character segment, ranks the recognition results in order from the highest similarity, and determines the highest similarity. The high recognition result is displayed here as the first recognition result.

Next, in step S3, the handwriting newly input by the user is received. Then, in step S4, it is determined whether or not the handwriting received in step S3 is an instruction to correct the character segmentation. If it is not an instruction to correct the character segmentation, processing other than the processing shown in FIG. 2 is performed, so the processing shown in this flowchart is ended. If the input is an instruction to modify the character segmentation, the process proceeds from step S4 to step S5.

In the case of this embodiment, as shown in FIGS. 3 to 10, the correction of the character segmentation is performed by enlarging the handwriting range in the character segmentation (FIG. 3) and reducing the handwriting range in the character segmentation (FIG. 4). , Erase part of handwriting in character segmentation (Fig. 5), move position of subset of handwriting in character segmentation (Fig. 6), increase size of handwriting subset in character segmentation (Fig. 7), character Reducing the size of the handwriting subset in the segmentation (Fig. 8), adding new handwriting in the character segmentation (Fig. 9), or transforming the shape of the handwriting subset in the character segmentation (Fig. 10). It will be either.

The details of each process will be described later with reference to FIGS. 11 to 29, but for ease of understanding, eight types of modification examples of character segmentation will be briefly described here.

As shown in FIG. 3, the expansion of the handwriting range in the character segmentation can be performed, for example, in Hiragana with "Ishii".
In the case where the characters are input by handwriting and the input of the first stroke and the second stroke of “i” that should be recognized as one character are divided into separate character segmentations and are recognized, The first stroke and the second stroke are modified so as to be included in the same character segmentation.

To reduce the handwriting range in the character segmentation, as shown in FIG. 4, for example, by inputting the kanji of "Yaguchi Watari", the characters "arrow" and "mouth" which should originally be recognized as two independent characters are used. "Is recognized as the character" knowledge "within one character segmentation,
Among the characters "knowledge", in order to recognize "arrow" and "mouth" as independent characters, the character segmentation is corrected into two.

As shown in FIG. 5, when the character “large” should be input but the character “heaven” has been input, the part of the handwriting in the character segmentation is deleted. It is used to erase one stroke and change it to the character "Large".

The movement of the position of the subset of the handwriting in the character segmentation is performed, for example, as shown in FIG. 6, when the character "Kou" is to be input, the second stroke of the character is projected above the first stroke. Since I typed it in, the letters "Sat"
When it is recognized that the first image is moved upward, the first image is moved to a position in contact with the second image.

To increase the size of the handwriting subset in the character segmentation, for example, as shown in FIG. 7, where "tai" should be entered in hiragana, the size of hiragana "i" is If the Hiragana "i" is recognized as a lower case "i" because it is smaller than "", change the lower case "i" to the normal size "i". It is used when

To reduce the size of the subset of handwriting in the character segmentation, for example, as shown in FIG. 8, where "2 places" should be input, the katakana "K" of the center character is greatly input. When this character is recognized as a normal size "ka" because it has been done, it is used when changing it to a lower case "ka".

As shown in FIG. 9, when a new handwriting is added to the character segmentation, the character "dog" should be originally input, but the character "large" is mistakenly input. If you notice from, add a point to the character segmentation including the character "Large",
This is used when changing this to the character "dog".

As shown in FIG. 10, the deformation of the shape of the handwriting subset in the character segmentation is such that the input is recognized as the alphabetic character "T" when the original katakana "A" should be input. In this case, the shape is deformed so that the right end portion of the first image is bent downward and the recognition result is used as "A" of katakana.

Returning to FIG. 2, when it is determined in step S4 that the input is an instruction to modify the character segmentation, the process proceeds to step S5, in which there is a recognition result matching the modification instruction. To determine. That is, at this time, the CPU 1 determines whether or not the recognition result corresponding to the corrected character segmentation is stored in the recognition result database DB2, and if such a recognition result has already been obtained, proceeds to step S6. Then, the recognition result is read out and displayed on the display unit 13.

For example, when the Chinese character "Yaguchi" is input as in the embodiment shown in FIG. 4, even if "knowledge" is obtained as the first recognition result, for example, as the second recognition result. If “Yaguchi” has already been obtained, if the instruction to reduce the range of handwriting in the character segmentation shown in FIG. 4 is made, the already obtained second candidate recognition result “Yaguchi” will be displayed.

On the other hand, if it is determined in step S5 that there is no recognition result that matches the correction instruction, the process proceeds to step S7, and handwritten handwriting having character segmentation that matches the correction instruction is generated. That is, at this time, the CPU 1 re-executes the segmentation of the character segmentation corresponding to the correction instruction, reads the handwriting corresponding to the corrected character segmentation from the handwriting handwriting database DB1, and sets it as the handwriting included in the character segmentation. Then, this is supplied to and stored in the correction handwriting handwriting database DB3 of the RAM3. The more detailed processing of this step S7 will be described later with reference to FIG.

Next, in step S8, the character segmentation stored in the correction handwriting handwriting database DB3 is recognized. Then, the recognition result obtained as a result of recognition is used as a correction recognition result database DB4 in the RAM3.
Supply to and store.

Next, in step S9, the contents of the recognition result database for correction DB4 are changed to the recognition result database DB4.
Rewrite the contents of 2. Then, after step S9, the process returns to step S2 to repeat the subsequent processes.

In this embodiment, step S3
In the above, the instruction to correct the character segmentation is received as a gesture (handwriting input), but it is also possible to make it input from the icon displayed on the display unit 13 or to make a predetermined input from the keyboard, mouse, etc. Is.

Further, although the handwriting handwriting database DB1 and the correction handwriting handwriting database DB3 are configured as independent ones, if the handwriting handwriting database DB1 is directly changed when the corrected handwriting handwriting is generated in step S7. The correction handwriting handwriting database DB3 can be omitted. In this case, the character recognition in step S8 is performed using the handwriting handwriting database DB1.

Similarly, the recognition result database DB2 and the correction recognition result database DB4 can be integrated into one. In this case, the recognition result database DB2 is directly changed as a result of the character recognition in step S8.

Here, the data structure will be described. The input handwriting is expressed as follows for each stroke. First image s [1] = {(x [1], y [1]), ..., (x
[N], y [n])} Second screen s [2] = {(x [n + 1], y [n + 1]),
・ ・ ・} Third image s [3] = {...} ... Kth image s [k] = {...}

That is, the handwriting data s [i] of the i-th stroke is represented by a sequence of coordinate points (x [a], y [a]), ..., (x
[B], y [b]).

Further, the character recognition result is expressed as follows. 1st place r [1] = {<s [a11], s [b11]: code [1] [1]>, <s [a12], s [b12]: code [1] [2]>, < s [a13], s [b13]: code [1] [3]>, ..., <s [a1q], s [b1q]: code [1] [q]>} 2nd place r [ 2] = {<s [a21], s [b21]: code [1] [1]>, <s [a22], s [b22]: code [1] [2]>, <s [a23], s [b23]: code [1] [3]>, ..., <s [a2p], s [b2p]: code [1] [p]>} ,.

The recognition result of the jth position is the start stroke s [m] and end stroke s [n] of the strokes (one stroke of handwriting) that form a character, and the character code code [] of the recognition result for the stroke group. The three elements of [] are set as one unit, and units for the number of character segmentations are provided. For example, in the second example from the top of FIG. Becomes

The process of generating the handwritten handwriting having the character segmentation matching the correction instruction in step S7 will be described below with reference to the second example from the top of FIG. That is, the handwriting sets of the first to fifth strokes are s [1], s [2], s [3], s
It is represented by [4] and s [5]. On the other hand, the character segmentation result G [i] is expressed as follows. G [1] = {s [1]} G [2] = {s [2]} G [3] = {s [3]} G [4] = {s [4], s [5]}

This character segmentation result G [i]
The character recognition result R [i] corresponding to is represented as follows. R [1] = {'s'} R [2] = {'1'} R [3] = {'s'} R [4] = {'s'}

In the case of the third processing from the top in FIG. 3, G
Since it is the process of combining [1] and G [2], the character segmentation G '[1] after the handwriting range expansion symbol is input.
Is as follows: G '[1] = {s [1], s [2]}

Then, s [1] and s [2], which are handwritings different from those before the input of the expansion instruction symbol, are stored in the correction handwriting handwriting database DB3.

The following is an example of handwriting data generated in response to the above-mentioned correction instruction.

-Expansion of handwriting range divided by character segmentation G [1] = {s [1]} G [2] = {s [2]} → G '[1] = {s [1], s [2]} Before inputting correction instruction After inputting correction instruction

Reduction of handwriting range divided by character segmentation G [1] = {s [1], s [2], s [3]} → G ′ [1] = {s [1]}

Addition of handwriting to character-segmented handwriting G [1] = {s [1], s [2], s [3]} → G ′ [1] = {s [1], s [ 2], s [3], s'}

Partial erasure of handwriting segmented by characters G [1] = {s [1], s [2], s [3]} G [2] = {s [4], s [5]} → G '[1] = {s [1]} G' [2] = {s [4], s [5]}

-Movement of position of character-segmented handwriting subset-Shape deformation of character-segmented handwriting subset-Expansion of size of character-segmented handwriting subset-Character-segmented handwriting subset Size reduction Above, the four instructions are common and are as follows. G [1] = {s [1], s [2], s [3]} G [2] = {s [4], s [5]} → G '[1] = {s' [1] , S '[2], s' [3]} G '[2] = {s' [4], s '[5]} where s' [i] is the internal data of s [i] ( It is the handwriting data whose point sequence coordinates, etc.) have been changed.

In the step S8, in the case of the embodiment of FIG. 3, the recognition result R'for the handwriting {s [1], s [2]} corresponding to the corrected character segmentation.
[1] becomes R '[1] = {' I '}.

Further, in step S9, in the case of the embodiment of FIG. 3, the character recognition result R [i] and the character segmentation result G [i] are as follows. R [1] = {'shi'} R [1] = {s [1]} R [2] = {'1'} R [2] = {s [2]} R [3] = {'shi '} R [3] = {s [3]} R [4] = {' I '} R [4] = {s [4], s [5]}

This is R '[1] = {' i '}, G'.
It will be replaced by [1] = {s [1], s [2]}. That is, do the following: R [1], R [2] → R ′ [1] G [1], G [2] → G ′ [1] R [3] → R [2] G [3] → G [2] R [ 4] → R [3] G [3] → G [2]

Next, a more detailed process of generating handwritten handwriting having character segmentation that matches the correction instruction in step S7 of FIG. 2 will be described with reference to the flowchart of FIG.

First, in step S21, if the handwriting data for correction DB3 does not contain the handwriting data necessary for character recognition in step S8 of FIG.
This is read from the handwritten handwriting database DB1 and stored. That is, since the character recognition in step S8 is executed based on the database stored in the correction handwriting handwriting database DB3, all necessary handwriting data must be prepared in the correction handwriting handwriting database DB3. There is.

Next, in steps S22 to S29,
It is determined whether the handwriting data (gesture) received in step S3 of FIG. 2 is one of the above eight types of correction instructions. If it is determined that none of the eight types of correction instructions is given, the process proceeds to step S30, and error processing is executed.

When it is determined in any one of the eight types of correction instructions in steps S22 to S29, the process proceeds to the corresponding steps S32 to S39, and the process corresponding to each type of correction instruction is executed. To be done.

That is, when it is determined in step S22 that the correction instruction is the handwriting range expansion instruction, step S32
At, the handwriting range expansion processing is executed. Step S
When it is determined in 23 that the handwriting range reduction instruction is given, the handwriting range reduction processing is executed in step S33.
When it is determined in step S24 that the handwriting part erasing instruction is given, a handwriting part erasing process is executed in step S34.

When it is determined in step S25 that the instruction is to move the position of a subset of handwriting, in step S35 a partial handwriting position moving process is executed. Step S2
When it is determined in 6 that the instruction is to increase the size of the handwriting subset, the partial handwriting size expansion process is executed in step S36. When it is determined in step S27 that the instruction is to reduce the size of the handwriting subset, step S3
In 7, the partial handwriting size reduction process is executed.

Further, when it is determined in step S28 that the instruction is a handwriting addition instruction to the handwriting, a handwriting addition processing to the handwriting is executed in step S38. When it is determined in step S29 that the instruction is a shape modification instruction of the handwriting, a shape modification process of the handwriting is executed in step S39.

Next, details of each processing of steps S32 to S39 will be sequentially described.

FIG. 12 shows details of the handwriting range expanding process in step S32. First, in step S51, the circumscribed quadrangle of the handwriting range expansion instruction symbol 51 is obtained. Next, the process proceeds to step S52, and a set of handwriting in which all or part of the circumscribed quadrangle obtained in step S51 is obtained. Then, in step S53, the range expansion code and the starting image number and ending image number of the handwriting obtained in step S52 are output to the correction handwriting handwriting database DB3 as the correction type code.

That is, as shown in FIG. 13, when the first stroke and the second stroke of the hiragana "i" are classified as different character segmentations, a range expansion instruction is given so that these two character segmentations are involved. When the symbol 51 is input, a circumscribed quadrangle of this symbol having a substantially V-shape facing in the lateral direction is obtained (step S51), and the handwriting data of character segmentation including all or a part of the circumscribed quadrangle is 1 It is put together as one character segmentation (step S52, S53).
This is recognized as the hiragana "i" in the process of step S8.

FIG. 14 shows the details of the handwriting range reduction processing in step S33. First, in step S61, the circumscribed quadrangle of the handwriting range reduction instruction symbol 52 is obtained. In step S62, the handwriting existing on the left side of the handwriting range reduction instruction symbol 52 is obtained in the character segmentation in which the handwriting range reduction instruction symbol 52 exists, and the start image number and the end image number of the handwriting are corrected by handwriting. It is supplied to and stored in the handwriting database DB3.

Further proceeding to step S63, in the character segmentation in which the handwriting range reduction instruction symbol 52 exists, the start image number and the end image number of the handwriting existing on the right side of the handwriting range reduction instruction symbol 52 are the correction handwriting handwriting database. It is supplied to and stored in DB3.

For example, as shown in FIG. 15, when the kanji "word" is input in one character segmentation and it is desired to recognize it as two characters instead of as one character, this Among the handwritings forming the “word”, a handwriting range reduction instruction symbol 52 having a substantially V-shape is input between the handwriting of “word” and the handwriting of “go”. Then, the circumscribed quadrangle of this symbol is obtained (step S61), and the handwriting (“word”) located on the left side of this circumscribing quadrangle and the handwriting (“go”) located on the right side of the circumscribed quadrangle are regarded as individual character segmentations. (Steps S62 and S63). Therefore, if the corrected character segmentation is processed in step S8 of FIG.
"Koto" and "go" are recognized as two independent characters.

FIG. 16 shows details of the handwriting part erasing process in step S34. First, in step S71, the circumscribed quadrangle of the handwriting part deletion symbol 53 is obtained.
Next, in step S72, a set of handwriting included in the circumscribed quadrangle obtained in step S71 is obtained. Also,
In step S73, the character segmentation information (handwriting) excluding the handwriting obtained in step S72 is generated from the currently displayed character segmentation, and this is output to the correction handwriting handwriting database DB2 as the correction handwriting handwriting data. , Remembered. Further, the process proceeds to step S74, and the handwriting (handwriting included in the circumscribed rectangle of the handwriting part erasing symbol 53) obtained in step S72 is erased from the correction handwriting handwriting database DB3.

For example, as shown in FIG. 17, in a state in which the character "heaven" is input by handwriting, the handwriting part erasing symbol 53 in the shape of the substantially alphabetic character "N" with respect to the first image of "heaven". If you enter, this handwriting part erase symbol 5
The handwriting of the first stroke included in the circumscribed quadrangle 3 is selected as the handwriting to be erased (steps S71 and S72). And this handwriting is the handwriting handwriting database DB3 for correction.
Character segmentation including the remaining handwriting of the second stroke to the fourth stroke is generated (step S7).
3, S74). As a result, when the character is recognized in step S8, the recognition result as the character "large" is obtained.

FIG. 18 shows the details of the partial handwriting position movement processing in step S35. First step S
At 81, the circumscribed quadrangle of the loop portion of the partial handwriting position movement symbol 54 is obtained. That is, the partial handwriting position movement symbol 54 is composed of a circular loop portion and a linear tail portion, and the circumscribed quadrangle of the loop portion is obtained. Next, in step S82, a set of handwriting included in this circumscribed quadrangle is obtained.

Further, in step S83, the starting point (X ₀ , Y ₀ ) of the tail portion of the partial handwriting position movement symbol 54 is obtained.
And the end point (X ₁ , Y ₁ ) is obtained. Next in step S84
The process proceeds, the coordinate values of the handwriting contained in the circumscribed rectangle obtained in step S82, the added _{(X 1 -X 0, Y 1} -Y 0). That is, data is generated by moving the handwriting in the direction of the tail of the partial handwriting position movement symbol 54 by the length thereof.
Then, the information on the character segmentation including the changed handwriting is output to the correction handwriting handwriting database DB3.

For example, as shown in FIG. 19, when the character "Sat" is being input, the first "Sat"
When the partial handwriting position movement symbol 54 is input so that a part of the image is included in the loop portion, the partial handwriting position movement symbol 5
The circumscribed quadrangle of the loop part of No. 4 is obtained (step S8).
1) Further, the handwriting (first stroke) within the circumscribed quadrangle is obtained (step S82).

Then, the coordinates of the start point and the end point of the tail portion are obtained (step S83), and the handwriting in which the handwriting of the first stroke is moved by the length in the direction of the tail portion is generated (step S83). S84). As a result, like this,
When the set of handwritings to which the position of the first stroke has been moved is subjected to character recognition in step S8, a recognition result of "work" is obtained.

Further, for example, as shown in FIG. 20, it is assumed that the user has handwrittenly input two characters of "tree" and "tree". Of the two strokes of the “tree”, the fourth stroke of the left “tree” character segmentation 1 does not originally intersect with the right stroke of the “tree” character segmentation 2. However, if the user
If the input is made so as to intersect the strokes, this input of one character segmentation is recognized as "bay".

Therefore, when the character segmentation is separated so as to be separated into two character segmentations, a handwriting corresponding to the separated character segmentation must be newly generated (the process in step S7 of FIG. 2 is performed. If not), when the left "tree" character segmentation 1 and the right "tree" character segmentation 2 are recognized independently of each other, the left character segmentation 1 is recognized as "tree", but the right side The character segmentation 2 of is the 4th part of the "tree" on the left side.
Since the image intersects the second image, there is a high possibility that it will be erroneously recognized as the character "book".

That is, when it is determined in step S5 of FIG. 2 that there is no recognition result matching the correction instruction, as shown in step S7, a handwritten handwriting having a character segmentation matching the correction instruction must be generated. It is difficult to recognize the recognized character "book" as the original character "tree" unless input is performed again.

However, in this embodiment, as shown in FIG.
As shown in FIG. 1, for example, when the partial handwriting position movement symbol 54 is input so that the first part to the fourth part of the tree on the left side are included in the loop part and the tail part moves to the left side, Since the intersection of the 4th stroke of the "tree" and the 2nd stroke of the "tree" on the right side is released, the character "tree" can be easily recognized even in the character segmentation 2.

FIG. 22 shows the details of the partial handwriting size enlargement processing of step S36 of FIG. First, in step S91, the circumscribed quadrangle of the rectangular loop portion of the partial handwriting enlarging instruction symbol 55 composed of the rectangular loop portion and the tail portion protruding to the outside of the rectangular loop is obtained. Then, in step S92, a set of handwriting included in the circumscribed quadrangle is obtained. Next in step S93
Then, the coordinates of the lower left point (X ₀ , Y ₀ ) and the upper right point (X ₁ , Y ₁ ) of the tail portion of the partial handwriting enlargement instruction symbol 55 are obtained.

Further, the process proceeds to step S94, and step S
Lower left vertex of a circumscribed rectangle of the square loop portion obtained in 91 (X _2, Y ₂₎ and the upper right apex (X _3, Y ₃₎ coordinate values are determined. Then, the process proceeds to step S95 and step S9.
The coordinate values of the handwriting (the handwriting included in the circumscribed quadrangle of the rectangular loop portion of the partial handwriting enlarging instruction symbol 55) obtained in 2 are calculated according to the following equations, and the corrected handwriting data is obtained.

[0086]

[Equation 1]

Here, x ', y'is the data of the handwriting after the correction, and (x, y) is the data before the correction.

As can be seen from this arithmetic expression, if the direction of the tail portion of the partial handwriting enlarging instruction symbol 55 is set to 45 degrees, the handwriting can be made without changing the ratio of the original handwriting in the vertical direction to the horizontal direction. It is enlarged, but if you set it to a direction other than 45 degrees,
The ratio will change and be expanded.

For example, as shown in FIG. 23, the hiragana "i"
Is recognized as a lowercase letter, when the partial handwriting enlarging instruction symbol 55 is input so as to enclose the character “i” in the rectangular loop portion, “i” is selected as the handwriting included in this rectangular loop portion ( In steps S91 and S92), the coordinates of this character "i" (lowercase) are changed according to the above equation (steps S94 and S95). Therefore, when the character is recognized in step S8, it is recognized as "i" (capital letter).

FIG. 24 shows details of the partial handwriting size reduction processing in step S37 of FIG. First, in step S101, the circumscribed quadrangle of the rectangular loop portion of the partial handwriting reduction instruction symbol 56 constituted by the rectangular loop portion and the tail portion projecting inward of the rectangular loop portion is obtained. Next, in step S102, a set of handwriting included in the circumscribed quadrangle is obtained. Then, in step S103, the lower left point (X ₀ , Y ₀ ) and the upper right point (X ₁ ,
The coordinates of Y ₁ ) are obtained.

Further proceeding to step S104, the lower left apex (X ₂ , Y ₂ ) and the upper right apex (X ₃ ,
The coordinate value of Y ₃ ) is obtained. Next, in step S105, the coordinate values of the handwriting (handwriting included in the circumscribing rectangle) obtained in step S102 are converted as correction handwriting data according to the following formula. In addition, this formula is the above-mentioned formula 1
However, since the coordinate values X ₂ and X ₃ , and Y ₂ and Y ₃ are different in size from the case of Expression 1, the calculation is performed to reduce the handwriting.

[0092]

[Equation 2]

For example, as shown in FIG. 25, when the katakana "ka" is recognized as a capital letter among the characters "2 places", the partial handwriting reduction instruction symbol 56 is changed to the character "ka" in the square loop portion. "" Is selected as the handwriting included in this rectangular loop portion (steps S101 and S102), and its coordinate values are converted according to the above equation (steps S103, S104, S1).
05). Therefore, the converted data is processed in step S8.
When you recognize the character with, you can get "K".

FIG. 26 shows details of the handwriting addition process to the handwriting in step S38 of FIG. First, in step S111, character segmentation in which the additional instruction handwriting 57 exists is obtained. Next step S1
Proceeding to step 12, the correction handwriting handwriting database DB corresponding to the character segmentation obtained in step S111.
The additional instruction handwriting is added to the end of the handwriting in 3. Then, in step S113, the starting image number and the ending image number corresponding to the character segmentation added with the handwriting data in step S112 are supplied to and stored in the correction handwriting handwriting database DB3.

For example, as shown in FIG. 27, when the character "large" is input and recognized, if the additional instruction handwriting 57 is added to the upper right of the first stroke of the character "large", the handwriting becomes " The characters are added to the three images forming "large" (steps S111 and S112) to form a character composed of a total of four images. As a result, this is step S
When the character recognition processing is performed in 8, "dog" is obtained as a recognition result.

FIG. 28 shows details of the shape modification process of the handwriting in step S39 of FIG. First step S
At 121, a handwriting to be combined with the input deformation instruction handwriting 58 is obtained. Next, proceeding to step S122, the coordinate point sequence data in the correction handwriting handwriting database DB3 corresponding to the handwriting obtained in step S121 is combined with the coordinate point sequence data of the input deformation instruction handwriting, and the correction handwriting is used. Supply to the handwriting database DB3.

For example, as shown in FIG. 29, when the alphabet letter "T" is input and displayed, if the deformation instruction handwriting 58 that bends to the lower right is added to the right end of the first stroke, it is added. The first stroke is selected as the handwriting (step S121). Then, the point sequence data of the deformation instruction handwriting 58 is added to the point sequence data of the first stroke (step S122). As a result, when the added data is subjected to character recognition processing in step S8, katakana "A" is obtained.

FIG. 30 shows an example of the format of the correction handwriting handwriting database DB3. As shown in the figure, first, data corresponding to each image is sequentially arranged. The image number of the image is placed at the beginning of the data of each image, and then the number n of coordinate points forming the image.
, And then, the coordinate values (x ₀ , y ₀ ), (x ₁ , y ₁ ), ..., (x _n-1 ,
y _n-1 ) are sequentially arranged. Then, the handwriting data section end symbol is arranged at the end of each of these image data.

Next to the handwriting data section end symbol, the correction type code and the number of character segmentations to be recognized again are arranged. The correction type code is a code for identifying which of the above-described eight types of correction processing. Next to the number of character segmentations to be re-recognized, the start image number and the end image number of the character segmentation to be re-recognized are sequentially arranged. And, the file end symbol is arranged at the very end.

In the above, characters are processed, but the present invention can also be applied to the case of processing graphics. 31 to 38 show an embodiment of this case. It should be noted that FIGS. 31 to 38 correspond to the respective embodiments of FIGS. 3 to 10 in the case of correcting the above-described eight kinds of characters.

In the embodiment shown in FIG. 31, as shown in FIG. 31A, three figures and a straight line connecting them are input. In contrast, as shown in FIG. The graphic segmentation is divided into six. Therefore, as shown in FIG. 7C, the handwriting range expansion instruction symbol 5
When 1 is input so that it first crosses two graphic segmentations, these 2 are segmented as shown in FIG.
One graphic segmentation is changed to one segmentation, and the handwriting that was previously placed in a separate graphic segmentation is treated as the handwriting in one segmentation. As a result, a correct recognition result (rectangle in this embodiment) is obtained.

In the embodiment shown in FIG. 32, as shown in FIG. 31A, as in the case of FIG. 31, three figures and line segments for connecting the figures are input, and these figures are shown in FIG. It is recognized as shown in (b). In this example, the first square and the next horizontal line segment are 1
Recognized as a shape within one shape segmentation. Therefore, when the handwriting range reduction instruction symbol 52 is input as shown in FIG. 7C, as shown in FIG.
The handwriting arranged on the left and right of the handwriting range reduction instruction symbol (gesture) 52 is processed as a graphic of each separate graphic segmentation.

In FIG. 33, as shown in FIG. 33 (a), three figures and horizontal line segments connecting them are input, and they are recognized as shown in FIG. 33 (b). ing. Of these figures, the figure on the far right has double input arc-shaped line segments on the left side, so to erase one of them, as shown in FIG. When the symbol 53 is input, the line segment corresponding to the handwriting part erasing symbol 53 is erased as shown in FIG.

In the embodiment of FIG. 34, as shown in FIG. 34A, a substantially quadrangular figure and one straight line extending downward from substantially the center of the figure are input. . Then, the figure is recognized as a figure in which a quadrangle and a straight line are combined, as shown in FIG. Therefore, as shown in FIG. 7C, when the partial handwriting position movement symbol 54 is made to include a part of a straight line in its loop portion and the tail portion is directed upwards for a predetermined length, As shown in FIG. 7D, this straight line is moved to a position where the square is divided into two.

In the embodiment of FIG. 35, an arrow is entered between a quadrangle and a triangle as shown in FIG. Then, as shown in FIG. 9B, among these figures, the triangular figure is recognized as a small triangle. In this state, as shown in FIG. 7C, when the partial handwriting enlargement instruction symbol 55 is input so that the triangle is included in the rectangular loop and the tail portion of a predetermined length is projected outward, This triangle is modified into a larger triangle as shown in d).

In the embodiment of FIG. 36, an arrow is entered between the circle and the quadrangle, as shown in FIG.
Then, as shown in FIG. 7B, the circle is recognized as a shape larger than the quadrangle. Therefore, as shown in FIG. 7C, when the partial handwriting reduction instruction symbol 56 is entered so that the square loop includes a circle and the tail portion projects inward, as shown in FIG. , This circle is reduced.

In the embodiment shown in FIG. 37, as shown in FIG. 37A, three figures and line segments connecting these are input, and as shown in FIG. Recognized as graphic segmentation. Then, when the additional handwriting 57 is input to connect the diagonal line of the leftmost quadrangle figure as shown in FIG. 7C, the diagonal line is added to this quadrangle as shown in FIG. The shape is recognized.

In the embodiment of FIG. 38, as shown in FIG. 38 (a), a figure in which a part of the line on the left side of the circle is missing is input, and this is an arc as shown in FIG. 38 (b). Is recognized as Therefore, as shown in FIG. 7C, the shape modification symbol 58 is formed so as to connect the missing parts of the circle.
When is input, the figure is recognized as a circle instead of an arc as shown in FIG.

Although the present invention has been described above by taking the case of processing characters and figures as an example, the present invention can be applied to the case of processing arbitrary shapes.

[0110]

As described above, according to the handwriting input display method of the present invention, when the intended recognition result cannot be obtained, the correction is instructed, and the handwriting having the segmentation matching the correction instruction is generated. When the correct recognition result is not obtained, it is possible to surely obtain the desired recognition result without re-inputting.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an embodiment of an apparatus to which a handwriting input display method of the present invention is applied.

FIG. 2 is a flowchart illustrating the operation of the embodiment of FIG.

FIG. 3 is a diagram illustrating expansion of a handwriting range within character segmentation.

FIG. 4 is a diagram illustrating reduction of a handwriting range within character segmentation.

FIG. 5 is a diagram illustrating partial deletion of handwriting in character segmentation.

FIG. 6 is a diagram for explaining the position movement of a subset of handwriting in character segmentation.

FIG. 7 is a diagram for explaining size enlargement of a subset of handwriting in character segmentation.

FIG. 8 is a diagram illustrating size reduction of a subset of handwriting in character segmentation.

FIG. 9 is a diagram illustrating addition of handwriting to handwriting in character segmentation.

FIG. 10 is a diagram illustrating shape deformation of a subset of handwriting in character segmentation.

FIG. 11 is a flowchart illustrating a more detailed process of step S7 of FIG.

12 is a flowchart illustrating a more detailed process of step S32 of FIG.

13 is a diagram illustrating the process of FIG.

14 is a flowchart illustrating more detailed processing of step S33 of FIG.

FIG. 15 is a diagram illustrating the process of FIG.

16 is a flowchart illustrating a more detailed process of step S34 of FIG.

FIG. 17 is a diagram illustrating the process of FIG. 16.

FIG. 18 is a flowchart illustrating a more detailed process of step S35 of FIG.

FIG. 19 is a diagram illustrating the process of FIG. 18.

FIG. 20 is another diagram illustrating the process of step S35 of FIG. 11.

FIG. 21 is still another diagram explaining the process of FIG. 35.

22 is a flowchart illustrating a more detailed process of step S36 of FIG.

FIG. 23 is a diagram illustrating the process of FIG. 22.

24 is a flowchart illustrating a more detailed process of step S37 of FIG.

FIG. 25 is a diagram illustrating the process of FIG. 24.

FIG. 26 is a flowchart illustrating a more detailed process of step S38 of FIG.

FIG. 27 is a diagram illustrating the process of FIG. 26.

28 is a flowchart illustrating a more detailed process of step S39 of FIG.

FIG. 29 is a diagram illustrating the process of FIG. 28.

FIG. 30 is a diagram illustrating a format of a correction handwriting handwriting database.

FIG. 31 is a diagram illustrating a process of expanding a handwriting range of a graphic.

FIG. 32 is a diagram illustrating a handwriting range reduction process of a graphic.

FIG. 33 is a diagram illustrating a handwriting part erasing process of a figure.

FIG. 34 is a diagram for explaining a partial handwriting position movement process of a figure.

FIG. 35 is a diagram illustrating a process of enlarging a partial handwriting size of a figure.

FIG. 36 is a diagram illustrating a partial handwriting size reduction process for a graphic.

FIG. 37 is a diagram illustrating a process of adding a handwriting to a handwriting of a figure.

FIG. 38 is a diagram illustrating a shape modification process of a handwriting of a figure.

FIG. 39 is a diagram illustrating a conventional character input method.

[Explanation of symbols]

 1 CPU 2 ROM 3 RAM 8 Bus line 11 Display controller 12 V-RAM 13 Display section 14 Tablet interface 15 Input tablet 16 Pen

Claims (3)

[Claims] 1. A handwriting input of handwriting, the input handwriting is stored, the input handwriting is divided into segmentations, the segmentation is recognized, and the result obtained by recognizing the input handwriting and the segmentation is obtained. When the recognition result is different from a predetermined one, the segmentation is changed, when the segmentation is changed, a predetermined handwriting is read out from the stored handwritings, and the change is made. A handwriting input display method characterized by newly generating the handwriting corresponding to the segmentation and recognizing the changed segmentation. 2. The handwriting input display method according to claim 1, wherein the change of the segmentation is performed by a gesture. 3. The handwriting input display method according to claim 1, wherein the change of the segmentation is any one of enlargement, reduction, deletion, movement, and modification of the input handwriting.