JPH09237321A - Device for recognizing handwritten character - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent the accuracy of character recognition from being reduced and to bring a handwritten character close to a free format by segmenting a stroke included in a character frame set up based upon the start point of a 1st stroke of an input character handwritten in a coordinate input means as one character. SOLUTION: A character frame display position calculating part 16 calculates the display position coordinate data of a character frame of one character based upon the coordinate data of the start point of the 1st stroke of the character out of stroke position information. A character segmenting part 18 judges the writing end of one character and outputs a character segmentation signal. A recognition processing part 22 compares the stroke shape feature information and stroke position information of strokes in a character frame segmented by the reception of the character segmentation signal with character shape feature information stored in a character dictionary part 20, executes character recognition and outputs a character recognition result. A display part 26 is superposed to the input face of a coordinate input device and displays the recognized character, the handwritten character and the character frame to be display data through the input face of the input device.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition apparatus suitable for use online, for example, to recognize a written input character in real time.

[0002]

2. Description of the Related Art Conventionally, in an online character recognition device for recognizing a written character in real time, an input character is input from a coordinate input device (hereinafter also referred to as a tablet).
In inputting a character, each stroke (one stroke) of an input character to be written is input as a coordinate data string of a position where the pen moves from pen-on to pen-off.

The following two methods are generally known for character input. One is a method (hereinafter, also referred to as a fixed frame method) in which input characters are written one by one in a character frame set on the tablet prior to character input. The other is a system without a character frame (hereinafter, also referred to as a character frame free system).

Character recognition of input characters is performed for each character. Therefore, the coordinate data of the input character is cut out for each character. In order to cut out each character, it is necessary to judge the end of writing one character. In determining the end of writing one character, generally, the following two methods are adopted as the following determination methods (1) and (2), respectively.

(1) In the case of the fixed character frame method described above, when a stroke is input from one character frame to another character frame, the stroke input to the character frame up to that point is output as one character. Alternatively, when the elapsed time from pen-up to the next pen-on becomes a certain value or more, it is judged that the writing of one character is finished, and the character is cut out.

(2) In the case of the character-frame-free system described above, it is assumed that the characters are separated from each other by a certain distance or more, so that when the strokes are separated from each other by a certain distance or more. , A method of judging that the writing of one character is finished is also used.

[0007]

However, in the conventional handwritten character recognition device, in the case of the fixed frame method, when writing the input character in the fixed character frame, the position and size of the input character are Be restricted. For this reason, there is a possibility that the character recognition rate may be deteriorated due to the character breakage caused by the difficulty of writing, and that if the stroke extends out of the fixed character frame, the character may be erroneously recognized. It was

On the other hand, when the character box is free, there is an advantage that it is not restricted by the character box, but there is a risk that the end of writing one character may be erroneously determined. For example, when deciding the end of a character based on the spacing between strokes, when characters such as kanji that are likely to separate radicals left, right, top, or bottom are written, or at close character spacing. When a character is written, it is difficult to correctly judge the end of writing one character. As a result, it is difficult to correctly cut out characters, and the accuracy of character recognition may be reduced.

For this reason, it has been desired to realize a character recognition device which can input characters in a character input environment closer to a free format without lowering the accuracy of character recognition.

[0010]

According to the character recognition device of the invention of this application, the character recognition device comprises a coordinate input device for outputting a coordinate data string for each stroke (one stroke) of a character handwritten on the input surface, A preprocessing unit for preprocessing the coordinate data string output from this coordinate input device is provided, and stroke position information and stroke shape characteristic information for each stroke are extracted from the coordinate data string processed by the preprocessing unit. A stroke data processing unit, and a character frame display position calculation unit for calculating display position coordinate data of the character frame of the one character based on the coordinate data of the start point of the first stroke of the one character among the stroke position information, and displaying A character slicing unit that inputs position coordinate data and stroke position information, and determines the end of writing one character and outputs a character slicing signal. Equipped with a character dictionary section storing character shape characteristic information, stroke shape characteristic information and stroke position information of strokes in a character frame cut out in response to a character cutout signal, and character shape stored in the character dictionary section. It is equipped with a recognition processing unit that performs character recognition by comparing with the characteristic information and outputs the character recognition result. The character recognition result, the coordinate data string and the display position coordinate data string of the character frame are input, and based on the coordinate data string. The handwriting character, the recognition character based on the character recognition result, and the character frame based on the display position coordinate are provided with a display processing unit that performs display processing and outputs display data so as to display the display data. The recognition character, the handwritten character and the character frame of the display data that overlaps the input surface of the input device are displayed via the input surface of this coordinate input device. Characterized by comprising comprises a radical 113.

Further, in the handwritten character input device of the present invention, preferably, an arbitrary size of the character frame is registered prior to the input of the handwritten character to the tablet, and the registered size is output to the character frame display position calculation unit. The character frame size registration processing unit is provided, and the selection switch for outputting the output of the stroke data processing unit to the character frame size registration processing unit or the recognition processing unit and the character frame display position calculation unit is provided. A cutout timer for measuring the elapsed time from the pen-off to the pen-on of the stroke may be provided for use in determining the end of writing one character in the copy.

Further, in the handwritten character input device equipped with the character box size registration processing section, more preferably, the character box having the size registered in the character box size registration processing section is:
When the character frame set based on the coordinates of the start point of the first stroke of one character and the stroke of the one character intersect, the size of the character frame is set so that the coordinates of the end point of the stroke are included within the character frame. It is desirable to include a character size correction unit that enlarges and corrects and outputs correction information to the character box size registration processing unit.

[0013]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of a character recognition device according to the present invention will be described below with reference to the drawings. It should be noted that the drawings to be referred to merely schematically show the sizes, shapes, and positional relationships of the respective constituent components to the extent that these inventions can be understood. Therefore, these inventions are not limited only to the illustrated examples.

(First Embodiment) A first embodiment of the handwritten character input device of the present invention will be described with reference to FIG. FIG. 3 is a functional block diagram provided for explaining the handwritten character input device of FIG. 1. The character recognition device of this embodiment is a coordinate input device (hereinafter, also referred to as a tablet) 1
0, a preprocessing unit 12, a stroke data processing unit 14, a character frame display position calculation unit 16, a character cutout unit 18, a character dictionary unit 20, a recognition processing unit 22, a display processing unit 24 and a display unit 26.

The tablet 10 outputs a coordinate data string for each stroke (one stroke) of the character handwritten on the input surface. In addition, the pre-processing unit 12 performs pre-processing on the coordinate data string output from the tablet 10. Here, as pre-processing, one or more processes such as noise data removal from the coordinate data sequence, smoothing of the coordinate data sequence, or detection of the start point of the coordinate data sequence and termination process are performed. Note that this end processing is processing for correcting that the end of the stroke becomes shorter if a filter is applied to the coordinate data string during noise data processing.

Further, the stroke data extraction unit 14
Extracts stroke position information and stroke shape characteristic information for each stroke from the coordinate data string processed by the preprocessing unit 12. This stroke position information can be represented by coordinate data of the start point and end point of the stroke, or coordinate data of the center point of the stroke, for example. Further, the stroke shape characteristic information is represented by being coded by being classified in the direction in which the stroke extends, for example. For coding, for example, (1) "|", (2)
The strokes are classified into code patterns of (1), (2), ... Such as “-”, (3) “/”, (4) “\”, ... And the shape feature of the stroke is expressed by this code. Then, the character of the character is extracted by combining the code (stroke shape characteristic information) of the stroke forming one character and the stroke position information.

Further, the character frame display position calculation unit 16 is
Among the stroke position information, the display position coordinate data of the character frame of the one character is calculated based on the coordinate data of the start point of the first stroke of the one character.

Further, the character slicing section 18 receives the display position coordinate data and the stroke position information, and judges the end of writing one character and outputs a character slicing signal. In this embodiment, when determining the end of writing one character, the writing is performed on the input surface of the tablet 10.
The coordinates of the starting point of the stroke after the first stroke are
When it is outside the character frame set based on the start point of the first stroke, it is determined that one character has been written from the first stroke to the stroke immediately before this stroke. Then, the character is cut out by cutting out the coordinate data string of the stroke in the character frame. Then, the next character frame is set based on the start point of the stroke outside the character frame.

Further, the recognition processing section 22 receives the stroke shape characteristic information and stroke position information of the stroke in the character frame cut out in response to the character cutting signal, and the character shape characteristic information stored in the character dictionary section 20. Are compared, character recognition is performed, and the character recognition result is output.

Further, the character recognition result, the coordinate data string, and the display position coordinate data string of the character frame are input to the display processing section 24. Then, the display processing unit 24 displays the handwritten character based on the coordinate data string, the recognized character based on the character recognition result, and the character frame based on the display position coordinate at a predetermined position. Output the processed display data. Here, the display processing is performed so that the recognized character is displayed at, for example, a position designated by a cursor or the like, or a center position of the erased handwritten character.

The display unit 26 overlaps the input surface of the coordinate input device, and displays the recognized characters, the handwritten characters and the character frame of the display data via the input surface of the coordinate input device.

Here, FIGS. 2A to 2D show display examples of character frames for handwritten input characters. In FIG.
The case where two characters "character" are input by handwriting is shown.

First, FIG. 2A shows a state where the start point of the first stroke is pen-on at the position indicated by S ₁ .
Then, the pen is moved in the arrow direction. Then, when the pen is turned on at this starting point, the character frame 28 calculated by the character frame display position calculation unit is immediately displayed on the display unit. Then, the display unit 26 displays the first character frame 28 overlaid on the handwriting of the handwritten character of the tablet. It is advisable to set the size of this character frame before inputting characters. Also, the relationship between the position of the start point and the display position of the character frame can be arbitrarily set prior to the character input, but here, the character is set so that the start point is located slightly above the center of the character frame. The frame is displayed.

Next, FIG. 2B shows a state in which one character of the first character "sentence" has been written.

Next, FIG. 2C shows a state where the start point of the next stroke is pen-on outside the first character frame corresponding to one character of the first character "sentence". When the first stroke of the second character is pen-on, this is detected by the character frame display position calculation unit, and the first character frame 2 is detected.
8 is erased and a second character frame 30 is newly set based on the coordinates of the start point of the first stroke of the second character (indicated by S ₂ in FIG. 2C). In FIG. 2C, the erased first character frame 28 is indicated by a dotted line.
As shown in FIG. 2C, the first character frame and the second character frame 30 may partially overlap with each other.

Next, FIG. 2D shows a state in which one character of the second character "character" has been written. Further, the end of writing the last character input by handwriting can be determined, for example, by the fact that the elapsed time in the pen-up state becomes a certain time or more. In addition, by pressing a dedicated key indicating the end of character input after the character has been written, the character slicing section can be made to determine the end of writing the character.

Although the second character is written on the right side of the first character in FIG. 2, the positional relationship of the second character with respect to the first character is not particularly limited.
The second character can be written at positions such as up, down, left, and right with respect to the first character.

As described above, according to the character recognition device of this embodiment, the first input character written on the coordinate input means is the first character.
Set the character frame based on the position of the start point of the stroke,
The stroke included in this character frame is cut out as one character. As a result, compared to the case of writing the input character in the preset character box, the restriction on the "position" of writing the input character can be relaxed, and as a single character compared to the case where the character box is free. The recognition range can be clarified. As a result, according to the character recognition device of the present invention, it is possible to provide an input environment closer to a free format without lowering the accuracy of character recognition.

(Second Embodiment) Next, a second embodiment of the handwritten character recognition apparatus of the present invention will be described with reference to FIG. FIG. 3 is a functional block diagram for explaining the handwritten character recognition device of this embodiment. still,
In FIG. 3, the same components as those in the first embodiment are designated by the same reference numerals as those in the first embodiment, and detailed description thereof will be omitted.

The handwritten character input device according to the second embodiment is provided with a character box size registration processing section 32 in which an arbitrary size of the character frame is registered prior to input of the handwritten character on the tablet. Then, the registered size is output to the character frame display position calculation unit.

Further, in this embodiment, the output of the stroke data processing section is output via the selection switch 34 to the character box size registration processing section 32 or the recognition processing section 22.
And is output to the character frame display position calculation unit 16. The selection switch 34 connects the stroke data processing unit 14 and the character frame size registration processing unit 32 only when registering the size of the character frame (the selection switch 34 is shown in FIG. 3).
The position is indicated by a. ). And at other times,
The selection switch 34 connects the stroke data processing unit 14 and the character frame display position calculation unit 16 (the selection switch 34 is at the position indicated by b in FIG. 3).

In addition, in this embodiment, in order to determine the end of writing of one character in the character slicing section,
A character cutout timer 36 is provided for measuring the elapsed time from the stroke pen-off to the pen-on.
The character cutout timer 36 is connected to the character cutout unit, and when the elapsed time from pen-off to pen-on exceeds the permissible time, the character cutout unit 36 issues information indicating the time-out to the character cutout unit 18. On the other hand, when the stroke is pen-on to the outside of the character frame, as in the case of the first embodiment, the character cutout unit 18 determines the end of writing the character, and the character cutout unit 18 outputs the character cutout timer 36. To send information to the timer to be initialized.

Next, with reference to FIG. 4, four examples of registration of the size of the character frame will be described. 4 (A)-
FIG. 3D is a diagram provided for explaining an example of a setting method for registering the size of the character frame.

(First Example) In registering the size of the character frame, first, the selection switch 34 is set to the side a in FIG. 3 to connect the stroke data processing unit 14 and the character frame size registration processing unit 32. To do.

Next, a character frame to be registered is input to the input surface of the tablet 10 by the method shown in FIG. 4 below, for example.

As shown in FIG. 4A, the tablet 1
If a perpendicular line 38 is written and input from 0 to the end point 38b at 0, the perpendicular line 3 represented by the X and Y coordinate systems in chronological order is displayed.
A coordinate data string of 8 coordinate points is output from the tablet 10 to the preprocessing unit 12. Then, in the preprocessing unit 12, similar to the case of character input, a predetermined preprocessing is performed on the coordinate data string. The coordinate data string subjected to the preprocessing is output to the display unit 26 via the display processing unit 24.
On the display unit 26, as shown in FIG. 4A, a vertical line 38 is displayed immediately below the handwriting on the input surface of the tablet 10 as a solid line corresponding to the handwriting.

Simultaneously with the display of the perpendicular line 38, the preprocessed coordinate data string is transferred to the stroke data processing unit 14.
Is input to In the stroke data processing unit 14, the position information of the starting point 38a and the end point 38b of the perpendicular line 38 is extracted as the stroke position information. This stroke position information is output to the character box size registration processing unit 32. In the character box size registration processing unit 32,
From the stroke position information of the perpendicular line 38, size data corresponding to a square character frame having the perpendicular line 38 as one side is calculated. If the end point 38b is further moved while pen-on is continued and the writing is continued, the size data of the character frame corresponding to the latest sampled coordinate data of the end point is successively recalculated and updated. Then, since the end point is fixed when the pen is turned off, the size data of the character frame is calculated and registered based on the coordinate data of the fixed end point.

Further, the size data that has been sequentially updated until the pen is turned off is sequentially input to the character frame display position calculation unit 16. Then, in the character frame display position calculation unit 16, while the selection switch is set to the side a and there is no direct input of data from the stroke data processing unit 14 to the character frame display position calculation unit 16, the size data is sequentially updated. Based on this, the position data matched with the position of the character frame having the perpendicular line 38 as one side is sequentially calculated. The sequentially calculated position data is input to the display unit 26 via the display processing unit 24. Then, the display unit 26 sequentially displays the character frame 40 as indicated by the broken line in FIG.

Further, the end point 38b shown in FIG.
It is a tentative point that has not been pen-offed yet. Therefore, in FIG. 4A, the character frame 40 is indicated by a broken line to indicate that the character frame is not yet determined. Then, after the end point is determined, the display processing unit 24 to which the final character frame size data registered in the character frame size registration processing unit 32 is input causes the display unit 26 to erase the perpendicular line 38 and to make the character frame solid line. Display it. By displaying the character frame with a solid line, it is indicated that the character frame size has been determined and registered.

Since the selection switch 34 is on the side a,
While the character frame is registered, the stroke data processing unit 1
4 to the recognition processing unit 22 and the character frame display position calculation unit 16
There is no data input to. Therefore, the character segment display information is not sent from the character frame display position calculation unit 16 to the character segmentation unit 18, and the recognition processing unit 22 does not perform character recognition. Therefore, the character recognition result is not sent to the display processing unit 24 and displayed on the display unit 26.

After the character frame is set, the selection switch 34 is set to the b side in FIG. 3 and handwritten character recognition is performed in the same manner as in the case of the first embodiment described above.

Here, "pen-on" means that the tip of a pen-shaped pointing tool is brought into contact with the input surface of the tablet, while "pen-off" is brought into contact with the input surface of the tablet. It means that the tip of the pointing device that was previously used is separated from this input surface.

(Second Example) Generally, it is difficult to input a complete perpendicular line 38 by handwriting as shown in FIG. Therefore, regarding the example of the registration of the size of the character frame when the written perpendicular becomes an oblique line, FIG.
This will be described with reference to FIG.

First, when a slant line 42 is written from the start point 42a to the end point 42b on the tablet 10, a coordinate data string of the coordinate points of the slant line 42 represented by the X and Y coordinate systems in chronological order is obtained from the tablet 10. It is output to the preprocessing unit 12. In addition, in FIG. 4B, a hatched arrow 42 indicates
The writing direction of the diagonal line 42 is shown.

Then, in the preprocessing unit 12, as in the case of character input, a predetermined preprocessing is performed on the coordinate data string. The coordinate data string that has been preprocessed is displayed by the display processing unit 2.
4 is output to the display unit 26. In the display unit 26,
As shown in FIG. 4B, a diagonal line 42 is formed immediately below the handwriting on the input surface of the tablet 10 as a solid line corresponding to the handwriting.
Is displayed.

Simultaneously with the display of the slanted line 42, the preprocessed coordinate data string is displayed in the stroke data processing unit 14.
Is input to In the stroke data processing unit 14, the position information of the diagonal line 42 such as the start point 42a and the end point 42b is extracted as the stroke position information. This stroke position information is output to the character box size registration processing unit 32. In the character box size registration processing unit 32,
From the stroke position information of the start point 42a and the end point 42b of the slanted line 42, size data corresponding to a square character frame 44 whose one side is the vertical component of the slanted line 42 is calculated.

By performing the same processing as in the case where the perpendicular line 38 is input, the vertical component of the slanted line 42 is set to one side, so that the size of the square character frame can be reduced without writing the complete perpendicular line. You can register.

(Third Example) Generally, the position of the starting point of the first stroke of the input character is not necessarily fixed with respect to the range of the entire input character. For example, "sentence"
Consider the case of inputting four characters, “letter”, “recognition” and “knowledge”. The starting point of the first stroke (first stroke) of "sentence" and "character" among these four characters is located above the center of the entire character, while the starting point of the first stroke of "recognition" and "sense" is , Is located to the upper left of the entire character. In addition, in hiragana, katakana, and kanji, in general, the starting point of the first stroke tends to be located above the center or above the left side of the entire character. Therefore, the position of the starting point of the first stroke with respect to the entire character range tends to have a greater variation in the horizontal direction than in the vertical direction.

Therefore, when the character frame is set based on the start point position of the first stroke, the shape of the character frame should be a horizontally long rectangle in order to ensure that the input character fits within the character frame. Is considered to be preferable.

Next, FIG. 4C shows an example in which the size of a horizontally long rectangular character frame is registered. In the case of the square character frame described above, the size of the character frame was registered with the vertical component of the vertical line or diagonal line as one side, but in the case of the rectangular character frame, the input diagonal line is used as the diagonal line of the character frame. To register.

Specifically, the starting point 46a is placed on the tablet 10.
When the diagonal line 46 is manually written from to the end point 46b, the coordinate data string of the coordinate points of the diagonal line 46 represented by the X and Y coordinate systems in time series is output from the tablet 10 to the preprocessing unit 12. In addition, in FIG. 4C, the arrow of the diagonal line 46 indicates the writing direction of the diagonal line 46.

Then, in the preprocessing unit 12, similar to the case of character input, a predetermined preprocessing is performed on the coordinate data string. The coordinate data string that has been preprocessed is displayed by the display processing unit 2.
4 is output to the display unit 26. In the display unit 26,
As shown in FIG. 4C, a diagonal line 46 is formed immediately below the handwriting on the input surface of the tablet 10 as a solid line corresponding to the handwriting.
Is displayed.

Simultaneously with the display of the slanted line 46, the preprocessed coordinate data string is displayed in the stroke data processing unit 14.
Is input to In the stroke data processing unit 14, the position information of, for example, the start point 46a and the end point 46b of the diagonal line 46 is extracted as stroke position information. This stroke position information is output to the character box size registration processing unit 32. Then, in the character frame size registration processing unit 32, a rectangular character having the vertical component of the diagonal line 46 as the vertical side and the horizontal component as the horizontal side is obtained from the stroke position information of the start point 46a and the end point 46b of the diagonal line 46. Frame 4
Size data corresponding to 8 is calculated.

Thereafter, by performing the same processing as in the case where the perpendicular line 38 is input, the size of the rectangular character frame whose diagonal line is the diagonal line 46 can be set and registered.

Although an example of a horizontally long rectangular character frame is shown in FIG. 4C, the size of a vertically long rectangular character frame whose diagonal line is the input diagonal line may be registered in the present invention. it can.

(Fourth Example) Next, with reference to FIG. 4D, an example in which the size of the character frame is set and registered after trial writing of the character will be described.

In setting the size of the character frame,
Before actually registering the size, if you actually try writing the characters on the tablet, you can see the size of the characters that are easy to write.
Then, the size of the character frame that is large enough to enclose the character of easy writing, for example, assuming a frame including the characters of trial writing, specify two points on the diagonal of the frame. You can set and register with.

Specifically, first, the selection switch 34 is set to the position c in FIG. By setting the selection switch 34 to this position, the stroke data processing unit 14, the character frame size registration processing unit 32, and the character frame display position calculation unit 16 are not directly connected. Therefore, even if a character is input to the tablet 10 for trial writing, the stroke position information of the trial writing character is not sent to the character size registration processing unit 32 and the character frame display position calculation unit 16.

Then, trial writing characters are input to the tablet 10 as in the case of normal character input. FIG.
In (D), the character "sentence" is input as the trial writing character. Then, the X and Y coordinate systems are displayed in chronological order,
A coordinate data string of trial writing characters is output from the tablet 10 to the preprocessing unit 12. In the preprocessing unit 12, similar to the case of character input, a predetermined preprocessing is performed on the coordinate data string. The coordinate data string subjected to the preprocessing is output to the display unit 26 via the display processing unit 24. Display 2
In FIG. 6, as shown in FIG. 4B, the test-written character is displayed immediately below the handwriting on the input surface of the tablet 10 as a solid line corresponding to the handwriting.

While the selection switch 34 is set to the position c, even if a trial writing character is input to the tablet 10, the stroke position information of this trial writing character is stored in the character size registration processing unit 32 and the stroke position information. It is not sent to the character frame display position calculation unit 16. Therefore, the character frame size is not registered and the character frame display position is not calculated. Further, since no data is input from the character frame display position calculation unit 16 to the character cutout unit 18, the character cutout unit 18 does not output the character cutout signal. Therefore, the character recognition unit 22
In, no character recognition processing is performed. As a result, the recognized character is not displayed on the display unit 26.

Next, the selection switch 34 is set to the position shown in FIG. By setting this position, the stroke data processing unit 14 and the character box size registration processing unit 32 are directly connected.

Then, here, the character frame is set by designating two points on the diagonal. for that reason,
First, assuming a character frame surrounding a trial writing character, the size of the character frame is set by designating two points on the diagonal of the character frame. When designating 2 points, tablet 10
On the upper left point of the trial writing character on the input surface of, ie, the upper left corner point (upper left point) 52 of the assumed character frame, pen on and off, and then the lower right point of the trial writing character, that is, the assumed Point at the lower right corner of the drawn text frame (lower right point) 5
Pen on and off at 4.

Then, the coordinate data string of the coordinate points represented by the X and Y coordinate systems of the upper left point 52 and the lower right point 54 are output from the tablet 10 to the preprocessing unit 12. Then, in the preprocessing unit 12, similar to the case of character input, a predetermined preprocessing is performed on the coordinate data string.
The pre-processed coordinate data string is input to the stroke data processing unit 14. The stroke data processing unit 14 extracts the position information of the upper left point 52 and the lower right point 54. This position information is input to the character box size registration processing unit 32 via the selection switch 34. In the character frame size registration processing unit 32, based on this position information,
Size data of a rectangular character frame whose vertical component is the vertical component of the diagonal line connecting the upper left point 52 and lower right component 54 and whose horizontal component is the horizontal component is calculated.

Since the selection switch 34 is set to the position a, the character frame display position calculation unit 16 is
No data is directly input from the stroke data processing unit 14. In this case, the character frame display position calculation unit 16 calculates position data that matches the position of the trial writing character “sentence” with the position shown in the character frame 56, and inputs the position data to the display processing unit 24. The display processing unit 24 causes the display unit 26 to display the character frame with a solid line. By displaying the character frame with a solid line,
It indicates that the character box size has been confirmed and registered.

(Third Embodiment) Next, a third embodiment of the handwritten character recognition apparatus of the present invention will be described with reference to FIG. FIG. 5 is a functional block diagram for explaining the handwritten character recognition device of this embodiment. Third
In the second embodiment, in addition to the configuration of the second embodiment described above, a character size correction unit 58 for correcting the character frame size is provided. In FIG. 5, components that are the same as those in the second embodiment are the same as those in the second embodiment.
The same reference numerals as those of the embodiment are given, and the detailed description thereof will be omitted.

Also in the third embodiment, the above-mentioned second
Similar to the above embodiment, the size of the character frame is registered in the character frame size registration processing unit 32. Then, when handwriting a character, a character frame of that size is set based on the coordinates of the starting point of the first stroke of one character.

By the way, when a character is input by handwriting, the stroke of the character may be out of the character frame. Therefore, when the character frame and the stroke of the one character intersect, the character frame size correction unit 58 enlarges and corrects the size of the character frame so that the coordinates of the end point of the stroke are included within the character frame. . The correction information of the enlarged and corrected character box is input to the character box size registration processing unit.

When a character frame of a size registered in the character frame size registration processing unit 32 and a character frame set based on the coordinates of the starting point of the first stroke of one character intersects the stroke of the one character In addition, the size of the character frame is enlarged and corrected so that the coordinates of the end point of the stroke are included within the character frame. The correction information of the enlarged and corrected character frame is fed back to and input to the character frame size registration processing unit 32. The character box size registration processing unit 32 to which the correction information is input corrects the size of the registered character box based on the correction information and re-registers the corrected size of the character box.

When it is not necessary to re-register the size of the character frame, the size of the character frame registered in advance may be used as it is for the subsequent character input. In such a case, for example, a changeover switch (not shown) is provided between the character box size correction unit 58 and the character size registration processing unit 32, and the switch is turned off so that the character box size correction unit 58 can change the character size. No feedback is given to the registration processing unit 32.

Next, with reference to FIG. 6, an example of correcting the size of the character box in the character box size correcting section will be described. 6A to 6C are diagrams for explaining the correction of the size of the character frame. Here, an example in the case of inputting the characters of “sentence” by handwriting will be described.

First, when the first stroke (first stroke) 60 of a sentence character is written on the input surface of the tablet 10, the character frame 62 is generated based on the position of the starting point S ₁ of the first stroke.
Is displayed. The size of the character frame 62 is set prior to character input and is registered in the character frame size registration processing unit 32.

Next, the second stroke (second stroke) 64 of the character "sentence" is written. In FIG. 6A, the case where the second stroke 64 protrudes from the character frame 62 is shown. In this case, the second stroke 64 intersects with the right side 62a of the character frame, and the end point E ₁ of the second stroke 64 is outside the character frame 62.

Next, by the character box size correction unit 58,
The character frame 62 is enlarged and corrected. Here, the right side 62a of the character frame intersecting the second stroke 64 is sequentially moved to the outside of the character frame so that the coordinates of the end point E ₁ are included within the character frame.

Here, as an example of the display method, the character frame 62 of the initially set size is displayed by a solid line, and the enlarged and corrected character frame portion 62b is shown by a broken line. By displaying in this way, it is shown that the second stroke 64 extends beyond the character frame 62 of the initially set size, and the range of character recognition is within the range of the enlarged and corrected character frame. I can show the writer what it is.

When the pen is turned off from the tablet at the end point E ₁ of the second stroke, the right side 62 of the initial character frame 62 is erased, and the enlarged and corrected character frame portion 62a is switched to the solid line display, thereby enlarging and correcting. The latter character frame 66 is shown.

Further, in this embodiment, the size of the character frame 66 after the enlargement and correction is fed back to the character frame size registration processing section 32 and input.

Next, as shown in FIG. 6B, the fourth stroke 68
Shows a case where the character is out of the character frame 66 once enlarged and corrected. In this case, the fourth stroke 68 and the lower side 66a of the character frame 66 intersect, and the end point E ₂ of the fourth stroke 66 is outside the character frame 66. Incidentally, FIG. 6 (B)
In FIG. 5, the erased right side 62a is indicated by a dashed line for explanation.

Next, by the character box size correction unit 58,
The character frame 66 is enlarged and corrected. Here, the lower side 66a of the character frame 66 that intersects the fourth stroke 68 is sequentially moved to the outside of the character frame so that the coordinates of the end point E ₂ are included within the character frame.

Here, as an example of the display method, the character frame 66 enlarged and corrected for the first time is displayed by a solid line, and the character frame portion 66b enlarged and corrected again is shown by a broken line. By displaying in this way, the fourth stroke 68 is
It is possible to show that the character frame 66 of the size once enlarged and corrected is protruded, and also to show the writer that the range of the character recognition is the range of the character frame again enlarged and corrected.

When the pen is off from the tablet at the end point E ₂ of the fourth stroke, the lower edge 66a of the character frame 66 is reached.
Is erased, and the enlarged and corrected character frame portion 66b is switched to the solid line display, so that the enlarged and corrected character frame 70 is displayed again. FIG. 6C shows a state in which the character frame 70 after being enlarged and corrected again is displayed by a solid line. However, FIG.
In (C), the erased lower side 66a is indicated by a dot-dash line for explanation.

Further, in this embodiment, the size of the character frame 70 after the enlargement and correction is again fed back to the character frame size registration processing section 32 and input.

According to the third embodiment, since the character box size correction section 58 is provided in addition to the character box size registration processing section 32, the stroke within one character is registered in the character box size registration processing section 32 in advance. Even if the character frame of the size that has been written out is written out, the character frame is enlarged and corrected according to the position of the end point of the stroke. Therefore, the stroke portion protruding from the character frame before the enlargement / correction is not truncated and is targeted for character recognition. As a result, even when the stroke is out of the original character frame, it is possible to secure the same character recognition rate as that when the character is written only within the range of the character frame.

Further, the size of the enlarged / corrected character frame is fed back and registered by the character frame size registration processing unit to register the enlarged / corrected character frame even for a character input after a character with a stroke protruding. Can be applied. As a result, it is possible to set a character frame having a more optimal size for the writer. Therefore, it is expected that the writer can be provided with a character recognition input environment with less restrictions and closer to a free format than in the fixed character frame method. Furthermore, since there are few restrictions, it is possible to expect an effect of improving the recognition rate by providing a more comfortable recognition input environment.

[0084]

According to the character recognition apparatus of the invention of this application, the character frame is set based on the position of the starting point of the first stroke of the input character written on the coordinate input means and included in this character frame. Cut out each stroke as one character. As a result, it is possible to relax the restrictions on the "position" of writing the input character compared to the case of writing the input character in a preset character frame, and to recognize it as one character compared to the case where the character frame is free. The range can be clarified. As a result, according to the character recognition device of the present invention, it is possible to provide an input environment that is less likely to reduce the accuracy of character recognition and is closer to a free format.

Further, if the function of registering an arbitrary size of the character frame is added by providing the character frame size registration processing unit, the restriction on the size of the input character can be relaxed. As a result, it is possible to provide an input environment closer to free format.

Further, by adding a function for correcting the size of the character frame by providing the character frame size correction unit, the restriction on the size of the input character can be relaxed. as a result,
It is possible to provide an input environment closer to free format.

[Brief description of drawings]

FIG. 1 is a block diagram for explaining a character recognition device according to a first embodiment.

FIGS. 2A to 2D are diagrams for explaining a method of setting a character frame in the first embodiment.

FIG. 3 is a block diagram for explaining a character recognition device according to a second embodiment.

4A to 4D are diagrams for explaining a method of setting a size of a character frame according to the second embodiment.

FIG. 5 is a block diagram for explaining a character recognition device according to a third embodiment.

FIGS. 6A to 6C are diagrams for explaining the correction of the size of the character frame in the third embodiment.

[Explanation of symbols]

 10: Tablet (coordinate input device) 12: Preprocessing unit 14: Stroke data processing unit 16: Character frame display position calculation unit 18: Character cutout unit 20: Character dictionary unit 22: Recognition processing unit 24: Display processing unit 26: Display Part 28: First character frame 30: Second character frame 32: Character frame size registration processing unit 34: Selection switch 36: Character cutout timer 38: Vertical line 38a: Start point 38b: End point 40: Character frame 42: Diagonal line 42a: Start point 42b: End point 44: Character frame 46: Diagonal line 46a: Start point 46b: End point 48: Character frame 50: Test writing character 52: Upper left point (upper left point) 54: Lower right point (lower right point) 56: Character frame 58: Character frame size correction unit 60: First stroke 62: Character frame 62a: Right side 62b: Enlarged and corrected character frame portion 64: Second stroke 66: 1 time) larger modified character frame 66a: lower side 66b: again enlarged modified character frame portion 68: 4-stroke 70: character frame is enlarged corrected again

Claims (3)

[Claims] 1. A pre-processing unit comprising a coordinate input device for outputting a coordinate data string for each stroke of a character handwritten on an input surface, the pre-processing unit performing pre-processing on the coordinate data string output from the coordinate input device. A stroke data processing unit for extracting stroke position information and stroke shape characteristic information for each stroke from the coordinate data string processed by the pre-processing unit; Based on the coordinate data of the starting point of, the character frame display position calculation unit for calculating the display position coordinate data of the character frame of the one character, the display position coordinate data and the stroke position information is input, and the one character The character extraction part that outputs the character extraction signal by judging the end of writing Comparing the stroke shape feature information and the stroke position information of the stroke in the character frame cut out in response to the character cutout signal, and the character shape feature information stored in the character dictionary unit. A recognition processing unit for performing character recognition to output a character recognition result, the character recognition result, the coordinate data string and the display position coordinate data string of the character frame are input, and handwriting based on the coordinate data string A coordinate processing unit configured to perform display processing so as to display a character, a recognized character based on the character recognition result, and a character frame based on the display position coordinate at a desired position, and output display data; Overlaps the input surface of the device, and
A handwritten character recognition device comprising a display unit for displaying recognized characters, handwritten characters and character frames of the display data via an input surface of the coordinate input device. 2. The handwritten character recognition device according to claim 1, wherein an arbitrary size of a character frame is registered prior to inputting a handwritten character on the tablet, and the registered size is used to calculate the character frame display position. For outputting to the character frame size registration processing unit, for outputting the output of the stroke data processing unit to the character frame size registration processing unit, or the recognition processing unit and the character frame display position calculation unit. A handwritten character, comprising a selection switch, and comprising a cut-out timer for measuring an elapsed time from the pen-off to the pen-on of the stroke, for use in judging the end of writing of one character in the character cut-out portion. Recognition device. 3. The handwritten character recognition device according to claim 2, wherein the character frame has a size registered in the character frame size registration processing unit and is set based on the coordinates of the starting point of the first stroke of one character. When the character frame and the stroke of the one character intersect, the character frame size is enlarged and corrected so that the coordinates of the end point of the stroke are included in the character frame, and the correction information is registered in the character frame size. A handwritten character recognition device comprising a character size correction unit for outputting to a processing unit.