JPH0738209B2 - Handwritten shorthand character recognition device - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a handwritten shorthand character recognition device for handwriting input of shorthand characters from a handwritten character input unit and recognizing the shorthand characters.

[Conventional technology]

A tablet and an input pen that draws characters on the drawing surface of the tablet are generally used for inputting handwritten characters.
This handwritten character input unit sequentially outputs data regarding the contact point with the taglet of the input pen tip, that is, the trajectory of the writing point by drawing a character on the drawing surface of the tablet with the input pen, and sends this to the character recognition unit. To do. The scribe is 1
Every time the writing of a character is completed, the character recognition unit is notified of the end. The character recognition unit responds to the writing end information and recognizes the handwritten character based on the data regarding the locus of the writing point.

As a reference for the technique related to input of handwritten characters, Japanese Patent Publication No. 57-6151, Japanese Patent Publication No. 19471/57,
Examples thereof include JP-B-57-16382, JP-A-55-61884, JP-A-53-45935 and JP-A-55-140975.

In addition, Japanese document creation apparatuses that process Japanese documents using computer devices have become widespread. This is for inputting Japanese "Yomi" with "Kana" or the like from a keyboard or the like, and converting this into a Kanji-Kana mix sentence. Among the above-mentioned handwritten character input devices, there are those that recognize not only English letters, numbers and Kana characters, but also Kanji characters, but in general, there are many that recognize English letters, numbers, and Kana characters. Development of a Japanese document creation device as an input device for "Yomi" is also in progress. That is, "reading" is input by the handwriting input unit, and the character recognition unit recognizes the "Yomi" and delivers it to the Kana-Kanji conversion device to convert it into a Kanji-Kana mixing document.

On the other hand, when handwriting a character, a shorthand character has been devised with the intention of performing it at high speed. It is said that by using this, it is possible to write about 150 to 350 characters per minute, although it depends on the degree of heat.

[Problems to be solved by the invention]

Considering the above-mentioned conventional technique, if a handwritten character input device is used and shorthand characters are used as the characters, it is possible to input "reading" at a higher speed, and it is possible to significantly improve the speed of creating a Japanese document. it can. However, when creating a document using shorthand characters, unlike the case of creating a document using general characters, it is rare to write shorthand characters one by one, and multiple characters are continuously separated. List without doing. Here, it is important to note that the shorthand text is continuously written in phrase units and the phrases are separated from each other. The continuity of this document is also a factor that further improves the writing speed. Also, as a feature of shorthand characters, the characters are one side except in special cases,
Therefore, it is possible to write with a single stroke, and this enables continuity of characters.

Therefore, in order to input this shorthand character by the handwritten character input means, it is necessary to make it possible to separate a plurality of characters described in succession. That is, if shorthand characters are used and the characters are input in units of one character, the shorthand effect will be lost.

The present invention has been made in view of the above points, and an object thereof is a handwritten shorthand character recognition device capable of separating shorthand characters continuously handwritten and capable of recognizing each shorthand character. Is to get.

[Means for solving problems]

As described above, when creating a document using shorthand characters, unlike the case of creating a document using general characters, it is rare to write shorthand characters one by one, and multiple characters are consecutive. Are described without separation. Each of the shorthand characters is represented by a gentle curve or a straight line for the purpose of writing the character quickly, and each character has a feature that it is distinguished by a sharp change in angle.

The present invention focuses on this point. When inputting a shorthand character using the handwritten character input means, the direction change angle of the writing point is monitored, and if this is at least a predetermined angle, the shorthand character It is possible to recognize each of the divisions by determining it.

That is, a feature of the present invention is that a handwriting shorthand character input unit that sequentially outputs data related to a locus of a writing point on a drawing surface, and data that is sequentially input from the handwriting shorthand character input unit, and based on the data, The direction change angle detection means for sequentially detecting the direction change angle of the locus of the writing point and the angle information from the direction change angle detecting means are sequentially input, and the position of the writing point is detected when the angle exceeds a predetermined angle. Is determined as a break point of the shorthand character, and the data from the handwritten shorthand character input means is sequentially input, and the position of the writing point is determined based on the data in response to the determination information from the determination means. A handwriting shorthand character recognition device comprising a recognition processing means for recognizing each shorthand character as a delimiter point.

[Action]

According to the above, when a handwritten shorthand character is handwritten on the drawing surface of the handwritten shorthand character input means, the handwritten shorthand character input means sequentially outputs data regarding the locus of the writing point. The direction change angle detection means sequentially inputs this data and sequentially detects the direction change angle of the locus of the writing point. When moving from one character to the next, the writer makes a sharp change in the direction of the writing point. When the direction change angle detecting means detects this steep angle, the judging means judges that the detected angle exceeds a predetermined angle, and judges the position of the writing point as a delimiter of the shorthand character. The recognition processing means sequentially inputs the data from the handwritten shorthand character input means and, in response to the judgment information from the judging means based on the data, recognizes each shorthand character with the position of the writing point as a delimiter point of the shorthand character. To do.

In this way, it is possible to obtain a handwritten shorthand character recognition device that can separate shorthand characters that are continuously handwritten and input and that can recognize each shorthand character.

〔Example〕

 An embodiment of the present invention shown in the drawings will be described below.

FIG. 2 is a block diagram showing the overall hardware configuration of a handwritten shorthand character recognition device to which the present invention is applied, which is of a type called a relatively small computer device, generally a personal computer device, or a business computer device. It is configured by connecting a coordinate input device as a handwritten shorthand character input device to the computer device. The coordinate input device 2a includes a tablet and an input pen that draws characters on the drawing surface of the tablet, as will be described later in the present embodiment. Reference numerals 2b, 2c, 2d and 2e are a central processing unit, a memory, a display unit and a keyboard which constitute the computer unit COM.

The coordinate input device 2a has a tablet TB and an input pen 3a for drawing a character on the drawing surface of the tablet TB, as shown in FIG. Therefore, the tablet TB according to the present embodiment is provided with a character input area 3b and areas 3d, 3e, 3c for designating various modes and the like on its drawing surface. The coordinate input device 2a has status information indicating whether or not the writing point, which is the tip of the input pen 3a, is in contact with the drawing surface of the tablet TB,
The X coordinate value information and the Y coordinate value information indicating the contact position of the writing point are sequentially output as needed. That is, when a character is handwritten in the character input area 3b with the input pen 3a, X and Y coordinate value information as data regarding a writing point is sequentially output. Various mode settings are instructed by pressing the input pen 3a against the corresponding areas 3d, 3e, 3c.

The memory 2c stores a shorthand character recognition means, a separation detection means, a kanji conversion processing means, and a program forming various means for effectively activating these means, and a central processing unit.
2b achieves a predetermined function by processing and executing the program. Further, the memory 2c provides various storage units for storing various data from the coordinate input device 2a and various intermediate data required for processing when the central processing unit 2b processes and executes each of these programs. In the case of the present embodiment, the display device 2d is used for such purposes as sequentially displaying the handwriting of the handwritten character being input or displaying the recognized character.

FIG. 1 shows the principle of the present invention, and is a block diagram showing the mutual relation of the above-mentioned means. Reference numeral 2a is a handwritten shorthand character input means, which is the coordinate input device shown in FIG. SHCR is a shorthand character recognition means, OCAD is a direction change angle detection means, and DCSN is a determination means. These means SHCR, OCAD, and DCSN are stored in the memory 2c in the form of a program as described above, and the central processing unit 2b processes and executes the program to achieve a predetermined function. In this figure, a coordinate input device 2a as a handwritten shorthand character input means sequentially outputs data relating to a tip contact portion of the input pen 3a with respect to the drawing surface 3b, that is, a locus of a writing point. The direction change angle detection means OCAD sequentially inputs the output data from the handwritten character input means 2a, and sequentially detects the direction change angle of the locus of the writing point based on the data. The determination unit DCSN sequentially inputs the direction change angle information from the direction change angle detection unit OCAD, and when the angle exceeds a predetermined angle, the position of the writing point is determined as the delimiter point of the shorthand character.

The shorthand character recognition means SHCR sequentially inputs the output data from the handwritten character input means 2a, and based on the data, in response to the judgment information from the judgment judgment means, the position of the writing point is used as a delimiter point of the shorthand characters. The recognition process of each shorthand character is executed.

Accordingly, it is possible to obtain a handwritten shorthand character recognizing device which can separate shorthand characters input by handwriting continuously and can recognize each shorthand character.

Hereinafter, each of these means will be described in detail. FIGS. 4 to 14 are flowcharts showing the configuration of programs constituting various processing means, and FIG. 15 is a memory configuration diagram showing the configuration of various storage units necessary for executing these respective means. . FIG. 16 is a diagram showing an example of shorthand characters in correspondence with hiragana characters.

FIG. 17 (a) shows the input pen 3a on the drawing surface 3b of the tablet TB.
Fig. 17 (b) shows the locus TR of "Haruga" written by the shorthand character in Fig. 17 (b). The locus of "Haruga" written by the input pen 3a on the drawing surface 3b of the tablet TB. It is the figure which showed TR.

18 to 27 are explanatory views for explaining the principle of recognition of shorthand characters according to this embodiment.

In describing each processing means in detail, first, FIG. 15 will be described in detail. This figure is a memory configuration diagram showing the configuration of the primary storage unit necessary for executing each processing means as described above, and each storage unit is set to a predetermined address of the memory 2c. The GDST is a coordinate value information acquisition state storage unit that stores an acquisition state of a signal received from the coordinate input device 2a, that is, a status signal, X coordinate value information, and Y coordinate value information.
STT, XDT, YDT are storage units for respectively storing the status information, the X coordinate value information, and the Y coordinate value information received from the coordinate input device 2a. PNF is the closest past input pen 3
The status information associated with the coordinate value of the writing point of a, that is, the previous pen state storage unit that stores the pen-up state or the pen-down state, and PST is the content of the current status information storage unit STT and the closest past From the contents of the status information storage unit PNF, the state of the input pen 3a, that is, when the pen is down,
It is a pen state storage unit that stores any state at the time of pen-up and pen-up.

The DOTS is a coordinate that sequentially stores status information, X coordinate value information, Y coordinate value information, and various information obtained by this, which are respectively received from the coordinate input device 2a, and sequentially and cyclically stores data regarding a locus of a writing point. It is a value storage unit. The coordinate value storage unit DOTS is logically configured in an annular shape, and cyclically stores information regarding the locus of the writing point. The STCM is a character code storage unit that sequentially stores and stores the stored character codes of shorthand characters.

Hereinafter, each figure will be described in detail with reference to FIG.

4 and 5 are flowcharts showing the main processing means, and particularly FIG. 5 shows the trajectory acquisition means TD for inputting various information output from the coordinate input device 2a to the computer device COM by using the interrupt function. It is a flowchart shown. The trajectory acquisition means TD first activates the coordinate acquisition means GD in step 5a as described later in detail, takes in the coordinate information of the writing point and the status information of the input pen 3a at that time, and stores this in the coordinate value storage unit. Sequentially store in DOTS. Continue, Step 5b
Then, as described later in detail, the pen state acquisition means PS is activated, the state of the input pen 3a is judged from the stored contents in step 5a, and this is similarly stored in the coordinate value storage unit DOTS.

In the main processing means shown in FIG. 4, first, in step 4a, the initialization means INIT is activated. This means INIT initializes various temporary storage units and each unit of the apparatus, as will be described later in detail. In the following step 4b, the character delimiter detection means SPDT is activated, as will be described later in detail. This means SPDT detects the delimiter of each shorthand character based on the data taken in by the trajectory acquisition means TD shown in FIG. That is, as shown in FIG. 17, the shorthand characters are composed of a plurality of characters written in one stroke, and therefore, when recognizing each character, the delimiter is detected to separate each shorthand character. In step 4b, the character delimiter detection means SPDT detects the delimiter of the shorthand characters, so that the shorthand character recognition processing means SHCR is activated in step 4c. This means SHCR responds to the character delimitation detection information from the character delimitation detection means SPDT and recognizes the shorthand characters based on the data for one shorthand character. Then, in step 4d, the character codes of the recognized characters are sequentially stored in the character code storage unit STCM. Subsequently, in step 4e, it is determined whether or not the input pen 3a is separated from the drawing surface 3b of the tablet TB. If it is not separated, the process is returned to step 4b again to recognize the next character. Here, the input pen 3a is the tablet TB
If it is far from the drawing surface 3b, the next character means NXCH, which will be described in detail later, is activated in step 4f. In the case of shorthand characters, one character is basically written in one stroke, but there are also two strokes for special characters such as semi-voiced sound, so this is determined. And if this is a semi-voiced sound, step 4g
Then, the process proceeds to step 4c, and character recognition is executed again.
If it is not a semi-voiced sound, move the process to step 4h to start the KJCN conversion processing means KJCN, and by that time the character code storage unit ST
Kanji conversion is executed based on the character code stored in the CM. Then, the converted characters are passed to a program such as a Japanese document creation device. In step 4i, it is determined whether or not to end the series of processes, and if not,
The processing is returned to step 4b, and the series of processing is repeated again.

By executing such a series of processes, the handwritten shorthand characters are converted into a kanji-kana mixed sentence.

The processing means shown in FIGS. 4 and 5 will be described in detail below.

FIG. 6 is a flow chart showing the details of the coordinate value acquisition means GD, the outline of which is indicated by 4a in the step of FIG. The means GD first receives various information from the coordinate input device 2a in step 6a by its activation, and then continues in step 6b.
In (4), which information the internal state of the means GD is in the input waiting state is identified. That is, the stored contents of the coordinate value acquisition state storage unit GDST are "0" for waiting status information, "1" for waiting X coordinate value information, and "2" for Y.
It is predetermined to wait for coordinate value information. Then step 6c,
In steps 6f and 6j, the data types of the input information and the waiting information are checked. In particular, when different types of input data are detected in step 6f and step 6j, the internal state is reset by clearing the storage section GDST in steps 6i and 6m. In addition, step 6d, step 6
In step 6k, the received information is stored in the corresponding storage unit ST.
The data is transferred to and stored in T, XDT, and YDT, and in steps 6e, 6h, and 61, the storage contents of the storage unit GDST are rewritten so that the waiting state is returned.

Then, finally, in step 6n, the storage units STT, XDT,
The status information and coordinate value information stored in the YDT are transferred and stored in the coordinate value storage unit DOTS.

That is, the coordinate value acquisition means GD sequentially acquires the information from the coordinate input means 2a in the order of the status information, the X coordinate value information, and the Y coordinate value information, and stores the acquired information in the corresponding storage units STT, XDT, YDT. By storing and receiving and storing all of this information, execution of subsequent processing means is prompted.

FIG. 7 is a flow chart showing details of the pen state acquisition means PS, the outline of which is shown in step 5b of FIG. Upon activation thereof, the means PS first refers to the stored contents of the status information storage unit STT in step 7a to determine whether the state of the input pen 3a is currently in the up state or the down state. Next, in steps 7b and 7c, the input pen 3
Detect the state change of a. That is, the current pen status,
That is, the stored contents of the status information storage unit STT and the previous status information storage unit PNF are compared, and if they are in the pen-down state and the pen-down state, respectively, the pen-down state is continuing, and if they are in the pen-down state and the pen-up state, the pen-down state and the pen-up state. If it is in the pen-down state, it is determined that the time is pen-up. Next, in step 7d, step 7e, and step 7f, information indicating the corresponding states is transferred and stored in the pen state storage unit PST. Further, when there is a change in the pen state, that is, when the pen-up state changes to the pen-down state or the pen-down state changes to the pen-up state, the current pen state is indicated in the corresponding steps 7g and 7h. The information is transferred and stored in the front pen state storage unit PNF in preparation for the next process. And
Finally, in step 7i, the contents of the storage unit PST are stored in the coordinate storage unit DOTS as described above.

By executing this pen state acquisition means PS, it is possible to identify whether the current pen state is at the time of pen down, is continuing pen down, or is the time of pen up.

As described above, the coordinates of the locus of the writing point by the coordinate acquisition means GD and the pen state acquisition means PS shown in FIGS. 6 and 7, and the input pen 3a.
The states of will be sequentially stored in the storage unit DOTS.

FIG. 8 is a flow chart showing the details of the initialization means INIT whose outline is shown in step 4a of FIG. The means
INIT sets the device, the temporary storage unit and the like to the initial state as described above. In this embodiment, the coordinate input device 2a
Is connected to the computer device COM via a communication port, the baud rate of this communication port is first set to 9600 [BPS] in step 8a.
Normally, the baud rate of the keyboard 2e is 9600 [BPS], so the keyboard 2e should be compatible with the keyboard 2e.
The same baud rate as 9600 [BPS], so keyboard 2e
Set the same baud rate as the keyboard 2e to maintain compatibility with. In the next step 8b, the keyboard 2e interrupt is disabled. According to the present embodiment, the Japanese can be input by the coordinate input device 2a instead of the keyboard 2e, however, when the keyboard 2e and the coordinate input device 2a are simultaneously input, a strange character not intended by the operator is provided. May be displayed on the screen, the hardware interrupt is disabled by executing this step 8b. However, there are times when the operator wants to input a special character key on the keyboard 2e, for example, "&", """,""","{","}". Since the special character keys are not recognized, a keyboard mode for playing the interrupt of the keyboard 2e is prepared. This switching selection key is prepared on the coordinate input device 2a. The operator selects this keyboard mode selection key. Press
It allows the hardware interrupt of the keyboard 2e and disables all the interrupts of the coordinate input device 2a. In addition, in order to transfer the interrupt from the keyboard 2e to the coordinate input device 2a, a processing program that recognizes that special key status information and key code has entered the keyboard buffer is arranged at the beginning of the keyboard interrupt. Remember. Next, in step 8c, the storage unit as various work areas is initialized, and in step 8d, the memory buffer for storing the recognized shorthand character code is initialized.

In the case of this embodiment, this is to secure 30 characters. Next, in step 8e, the position of the cursor on the display screen of the display device 2d is stored.

This completes the initialization of the area around the coordinate input device 2a in the present embodiment.

Next, a shorthand character recognition process will be described. In this embodiment, as the basic character recognition of shorthand characters, the angle of the shorthand characters, the distinction of the length of the shorthand characters, the relative angle of the shorthand characters, the irregularity ratio of the shorthand characters, and the pattern matching of the shorthand characters are adopted.

FIG. 9 shows the character delimiter detection means shown in step 4b of FIG.
It is a flowchart which shows the detail of SPDT. The means SPDT
Is a processing means for detecting the delimiter of each shorthand character as described above. The x-coordinates and the y-coordinates of the loci of the shorthand characters drawn by the writer on the drawing surface 3b of the tablet TB are sequentially stored in the coordinate value storage unit DOTS by the locus acquisition means TG shown in FIG. In step 9a, this coordinate value information is sequentially read, and the angle θi between the coordinate value (Xi, Yi) of the position where the writing point is present and the coordinate value (Xi _{+ 1} , Yi _{+ 1} ) of the next writing point Is calculated by the following equation (1) and stored. In addition, below,
This θi is referred to as a shorthand inter-dot vector.

Subsequently, in step 9b, the difference between the angle value θi calculated in step 9a and the selected angle value θ _i-1 is calculated, and the first value is calculated.
As shown in FIG. 7, a point SP where the vector θi between shorthand points has a rapid change is detected. In the case of the present embodiment, the abrupt change of the inter-shorthand point vector θi is 60 degrees or more. If there is no abrupt change in the inter-shorthand point vector θi in step 9b, the process is returned to step 9a, and the angle change with the next writing point is detected again. Then, if there is a sudden change in the vector θi between the shorthand points, this point is detected as the break point SP.
In this way, it becomes possible to separate one character from the shorthand inter-dot vector θi and the locus coordinates xi, yi of the writing point. As is clear from this, step 9a constitutes the direction change angle detection means OCAD, and step 9b constitutes the determination means DCSN. The start of writing can be detected by the pen-down information, and the end of writing can be detected by the pen-up information.

FIG. 10 is a flow chart showing its outline in step 4c of FIG. 4 and showing the details of the multi-speed written character recognition processing means SHCR.

First, the principle of angle recognition will be described. The angles in which the shorthand characters are written are as shown in Fig. 18. That is, depending on the angle, five types of lines are drawn. Set line A to 0 degrees, line B to 30 degrees, line C to 90 degrees, line D to 120 degrees, and line E to 1 degree.
50 degrees. Since a slight deviation occurs in the angle during writing, the angles + 5 ° and −5 ° of the lines A to E are recognized as the angles.

However, the points to be noted are the line D and the line E. When the writer writes the lines D and E in shorthand, this is difficult to distinguish by the mere angle. Therefore, the feature that line D is steep from top to bottom and line E is gentle from bottom to top is used to recognize the distinction between line D and line E. To do.

Basically, the shorthand characters are drawn at the above-mentioned five types of angles, and therefore the following two angles are used to recognize each shorthand character. Therefore, as its preprocessing, step 10
The sample point sequence of x, y coordinates for one character divided by a is N
This point is approximated to 6 points in this embodiment. 19th
FIG. 20 and FIG. 20 show an example of a shorthand character of “ha”, which is shown in the case of equal division approximation with 6 sample points.
In step 10b, the angle character recognition means AGRG for recognizing each shorthand character using the angle is activated. FIG. 11 is a flow chart showing the details of the angle character recognition means AGRG. First, in step 11a, the sum of the relative angles of the sample points is calculated. In FIG. 19, the angle θi (i = 1 to 6) is the starting point
The angle from Ps (xs, ys) to the sample point Pi (i = 1 to 6) is shown. This angle θi is calculated by the following equation (2).

Then, the sum of these relative angles will be described by the following equation (3).

Next, in step 11b, based on the calculated value, the type of angle of the shorthand character and the unevenness of the character are detected. FIG. 21 is a diagram in which each of the shorthand characters shown in FIG. 16 is classified according to the length, the angle type, and the unevenness, and the shorthand characters are recognized according to such classification. In this figure, the corresponding "Hiragana" of the shorthand characters are added in parentheses. Then, in step 11c, the sum of the difference degrees in the angle of the sample points is calculated. That is, the angle θ in FIG.
i (i = 1 to 6) is the difference between the vectors connecting the sample points, and in step 11c, the sum of the angles θi is calculated. The sum of the relative angles calculated as described above and the sum of the difference degrees of the angles are compared in step 11d with the corresponding angle values of the standard dictionary created in advance, and the characters with a small angle difference between the two are stepped. Select multiple items at 11e and memorize them.

Next, recognition of the length of shorthand characters will be described. Shorthand
The Japanese syllabary is as shown in FIG. As you can see from this figure, in shorthand, long lines are almost twice as long as short lines. In order to distinguish a long line from a short line, in this embodiment, short lines and short lines are usually distinguished based on 1.2 [cm]. Generally, long characters are 1 cm or more and average 1.5 cm, and short characters are
The average length is 7 [mm]. However, since the character size varies depending on the shorthand writer, the present embodiment is provided with a learning mode in which the writer can specify the length for distinguishing the length. In step 10c, it is determined whether or not this learning mode is designated. If yes, the learning means STUM is started in step 10d. This learning means STUM shows its detailed flowchart
As shown in the figure, in step 12a, specify the character whose sum of relative angles and the sum of differences in angle difference are similar, and ask the writer to write the character in a size that the writer can write on a daily basis. Prompt for input. Then, in step 12b, the written character is displayed on the display device 2d as it is, and the writer himself / herself recognizes the respective sizes at the same shorthand character angle. Then, when the writer writes characters with different sizes, such as the letters "a" and "ko", in reverse size,
This is determined in step 12c, and in step 12d the display device 2d
A message stating that there is a mistake is displayed on the screen, and re-input is requested. Then, by repeating these steps, the point that distinguishes the length of the handwritten character of the writer is recognized and stored in step 12e. The "character size" described above is obtained by the following method in this embodiment. That is,
As shown in Fig. 22, among the sample points recognized as one character, the barycentric coordinates (X, Y) thereof are calculated by the following formula.

The length obtained by doubling the distance from the center of gravity thus obtained to the sample point that is farthest away is defined as "character size".

In step 10e, the size of the shorthand character is recognized as described above, and in the subsequent step 10f, the size of the shorthand character, the sum of the relative angles described above, and the sum of the difference degrees of the angles described above are used as the shorthand characters. Is recognized and a plurality of candidate characters are selected. In step 10g, the character recognition means BRGM by the approximate direction vector series method is activated.

The means BRGM, as shown in FIG. 13 in detail, correctly recognizes a shorthand character from a plurality of selected candidate characters by the approximate direction vector series method of the following segments. As a pre-process, the means BRGM converts the sample point sequence data into data of the standard dictionary size of candidate characters. Then, the calculation based on the approximate direction vector series method shown below is executed.

The approximate direction vector is the difference value itself between the end points of the segment. P _{1 is} the feature point of (N + 1) that divides one character into N equal parts.
If (start point) to P _{N + 1} (end point), the approximate direction vector Δ
Pi _{is, ΔPi = Pi +1 -Pi = (} Xi +1 -Xi, yi +1 -yi) However, i = 1~N, Pn = ( xn, yn), n = 1~N + 1
Is.

The stroke distance ds between the input character and the standard pattern is obtained as the sum of the distances between the characteristic parameters.

That is, Becomes In step 13b, the standard dictionary pattern of candidate characters obtained by the size and angle of the character and the shorthand characters of the writer are calculated by the above equation (7), and the character having the minimum parameter distance is selected. In step 13c, the selected character and the handwritten character of the writer are compared based on the distance between the approximate direction vectors obtained by the equation (6). Then, if the distance between the written character and the approximate direction vector of the standard dictionary is an approximate value, it is stored that the character is confirmed in step 13d,
If not, it is stored in step 13e that the character is not recognized.

Then, returning to FIG. 10 again, step 10h is executed.
This step 10h is a predetermined judgment as to whether or not a character is recognized, and more specifically, this is judged according to the stored contents at step 13d or step 13e in FIG.
Here, if the character has been recognized, the execution of the shorthand character recognition means SHRG shown in FIG. 10 ends. Here, if the character is not recognized, the special character recognition means SCR is executed in step 10i.
Start G. FIG. 14 is a flowchart showing details of the special character recognizing means SCRG. The means SCRG is dull,
Semi-voiced sounds, nasal sounds, and nasal voiced sounds are determined, and character recognition is executed according to these. That is, in step 14a, a dull sound, in step 14b, a dull sound, in step 14c, a nasal sound
In step 14d, the nasal sounds are judged.

First, an example of recognition of dull sounds will be described with reference to FIG. From this figure, it can be seen that adding a small circle to the end of the line produces a dull sound. Also, you can see that if the shorthand character written before is a straight line, this small circle is attached to the top or the left, and if it is a curve, all the small circles are attached to the inside. As will be described later, since the abbreviation of "i" is also a circle, in order to distinguish it from the circle, the size of the circle of the dakuon is, in the case of this embodiment, 1/5 or less of the long character of the Kiyon. I will recognize it as a dull sound. When there is no abrupt change in the angle in the equation and there is an intersection of the x-coordinate and the y-coordinate, and the circle including the intersection meets the conditions of the circle of dull sound described above, this is recognized as a dull sound. Then, in step 14e, the cursor position of the display device 2d and the character code of the preceding one character, Kiyone, are replaced with the character of the voiced sound, and the cursor is returned to the original position.

Next, the case of semi-voiced sound in step 14b will be described. 24th
As you can see from the figure, the written letters are short,
If it is a straight line and the distance from the last writing point one character before is a certain length or more, this is determined to be a semi-voiced sound, and step 14e is executed as described above.

Next, nasal sounds and nasal dull sounds will be described with reference to FIGS. 25 and 26. If you hit the end of the line to the upper right, this is recognized as "n". However, as shown in FIG. 27, distinction is made between straight lines and joints of straight lines, and between curves and joining of curves. Since the shorthand characters need to change their angle abruptly at the connection between the characters, if the lines are straight lines, shift the lines slightly, and if the lines are curves and curves and the lines are straight Stop the brush.

In this case, nasal letters are recognized when the length of the splashing is 1/2 or less and 1/5 or more of the length of the short letters of the clean sound, and when the length is 1/5 or less, the line Recognize this as a seam.

In the case of nasal voice, write the letter "U" above or to the left of the line when writing "n" in the voice as shown in Fig. 26. In this case, Kiyone is already recognized as Kiyone when writing "n". In order to distinguish the letters "o" and "ne" from each other, when the size of the letters "o" and "ne" is half or less, this is recognized as a nasal sound. Then, in step 14f, the character code of the pure sound described one before is corrected to the character code of the nasal noise, and the cursor is advanced one step.

When the shorthand character is recognized as described above, the character code is stored in the character code storage unit STCM as described in step 4d of FIG. Then step 4e above
Then, it is determined whether or not the input pen 3a is separated from the drawing surface 3b of the tablet TB based on the pen state information detected by the pen state acquisition unit PS shown in step 5b of FIG. If not separated, the process is repeated from step 4b, the next shorthand character is recognized, and the recognized character code is sequentially stored in the character code storage unit STCM in step 4b.

When it is detected in step 4e that the input pen 3a is separated from the drawing surface 3b of the tablet TB, the next character determination means NXCH in step 4F is activated. As described above, when the input pen 3a is separated from the drawing surface 3b, the shorthand character can be recognized as a segment break. However, when the character is the semi-voiced sound as shown in FIG. 24, the input pen 3a is separated from the tablet TB in the middle of one character.
Therefore, the next character determining means NXCH in step 4f determines the next coordinate data, and determines whether or not this forms semi-voiced sound together with the last recognized character. The means NXCH is substantially steps 14b and 14e in FIG. 14, and detailed description thereof will be omitted. If this forms a semi-voiced sound, the process moves to step 4c in step 4g, and the last voiced character is subjected to the semi-voiced sound process.

As described above, if the separation of the input pen 3a is detected in step 4e and the next character does not form a semi-voiced sound, the kanji conversion processing means KJCN is activated in step 4h. When the means KJCN is activated, the character code string stored in the character code storage unit STCM until then is delivered to the Kanji conversion processing means KJCN. Kanji conversion processing means
As the KJCN, the one that has been conventionally used in the Japanese document creation device can be used, and its details are omitted.

In the case of the present embodiment, the Kanji conversion processing means KJCN in the phrase conversion format is used. One of the reasons for this is that shorthand characters are written in phrase units. The other one is the conversion rate of kanji. That is, this is because the correctness rate of conversion in phrase units is extremely high as compared with the Kanji conversion process in sentence units.

As described above, according to the embodiment, the shorthand characters are handwritten on the drawing surface 3b of the tablet TB by the input pen 3a. Then, the trajectory acquisition means TG stores the coordinate information about the tip of the input pen 3a in contact with the drawing surface 3b, that is, the coordinate information about the trajectory of the writing point and the state information about the contact / separation of the input pen 3a with respect to the drawing surface 3b in the coordinate value storage unit DOTS. Are sequentially stored and stored in. The character delimiter detection means SPDT includes a direction change angle detection means OCAD and a determination means DCSN. The direction change angle detecting means OCAD sequentially reads the stored information from the coordinate value storage unit DOTS and sequentially detects the direction change angle of the locus of the writing point. The determination unit DCSN sequentially inputs the angle information from the direction change angle detection unit OCAD, and when the angle exceeds a predetermined angle, the position of the writing point is determined as the delimiter point of the shorthand character. The shorthand character recognition processing means SHCR recognizes the character for each character delimited by the character delimitation detection means SPDT and sequentially stores the recognized character codes in the character code storage unit STCM. When the handwriting of one segment is completed, the writer separates the input pen 3a from the drawing surface 3b in order to handwrite the next segment.

The separation detecting means PNUP constantly monitors the state of the input pen 3a, detects that the input pen 3a has separated from the drawing surface 3b, and outputs writing point separation information. The Kanji conversion processing means KJCN responds to this writing point separation information and converts it into a kanji-kana mixed sentence based on the character code string stored in the character code storage unit STCM up to that point.

As is clear from this, the drawing surface 3b of the tablet TB is
By handwriting the 1-segment, and when the input pen 3a is separated from the drawing surface 3b by the end, the kanji conversion processing is automatically executed.
That is, the writer does not need to operate the conversion instruction key or the like when converting the kanji. Moreover, if you do this,
A characteristic of shorthand characters, that is, a plurality of shorthand characters are continuously written in segment units, so that the target of Kanji conversion is segment units, and the conversion efficiency can be improved.

As described above, in the embodiment, the case where the tablet TB and the input pen 3a for drawing a character on the drawing surface of the tablet TB are used as the handwriting shorthand character input means has been described, but the present invention is not limited to this. Various coordinate input devices used as general handwritten character input means can be used.

Further, in the embodiment, the shorthand character confirmation means SHRG does not particularly mention this character code which finally recognizes the shorthand character as a character code, but this may be a corresponding kana character code, and It may be a shorthand character code assigned to each shorthand character. Especially in the case of kana character code, there is an advantage that the conventional kanji conversion processing means KJCN can be used almost as it is. This is
Once recognized by the shorthand character code, Kanji conversion processing means KJCN
If the character code is transferred to the kana character code, it can be converted into the kana character code, and the same advantages as described above can be obtained. In addition to this, a Kanji conversion dictionary for a shorthand character code may be provided.

〔The invention's effect〕

As is clear from the above description, according to the present invention, since the breakpoints of the shorthand characters that have been handwritten continuously can be automatically determined, it is possible to separate the handwritten characters that have been handwritten continuously. A handwritten shorthand character recognition device capable of recognizing characters can be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing the principle of the present invention, FIG. 2 is a block diagram showing the overall configuration of a handwriting shorthand character conversion device to which the present invention is applied, and FIG. 3 is an external view showing an example of a coordinate input device. 4 and 5 are flowcharts showing the main processing means, FIG. 6 is a flowchart showing the coordinate acquisition means, and FIG.
FIG. 8 is a flow chart showing the pen state acquisition means, FIG. 8 is a flow chart showing the initialization means, FIG. 9 is a flow chart showing the character division detection means, FIG. 10 is a flow chart showing the shorthand character recognition processing means, and FIG. FIG. 12 is a flow chart showing the angle character recognition means, FIG. 12 is a flow chart showing the learning means, FIG. 13 is a flow chart showing the character recognition means by the approximate direction vector series method, FIG. 14 is a flow chart showing the special character recognition means, and FIG. Is a memory block diagram showing a configuration of various storage units used by various processing means, FIG. 16 is a diagram showing an example of shorthand characters, FIG. 17 is a diagram showing an example of a locus of a writing point on a drawing surface, and 18 The figures are for explaining the angles of letters used for shorthand, Figure 19, Figure 20
Figure 21, Figure 21 (a), Figure 21 (b), Figure 22 is a diagram for explaining the principle of shorthand character recognition, Figure 23 is a figure showing an example of the characters of dakuon, and FIG. FIG. 25 is a diagram showing a character string of nasal sounds, FIG. 26 is a diagram showing a character string of nasal sounds, and FIG.
FIG. 27 is a diagram for explaining the principle of recognition of shorthand characters. 2a: Handwritten shorthand character input means, SHCR ... Shorthand character recognition means, OCAD ... Direction change angle detection means, DCSN ... Judgment means.

Claims (3)

[Claims] 1. A handwritten shorthand character input means for sequentially outputting data on a locus of a writing point on a drawing surface, and data from the handwriting shorthand character input means are sequentially input and the direction of the locus of the writing point is based on the data. Direction change angle detection means for sequentially detecting the change angle and angle information from the direction change angle detection means are sequentially input, and when the angle exceeds a predetermined angle, the position of the writing point is delimited by a shorthand character. Determining means and the data from the handwriting shorthand character input means are sequentially input, and in response to the determination information from the determining means based on the data, each shorthand character uses the position of the writing point as a delimiter point of the shorthand character. Handwriting shorthand character recognition device comprising a recognition processing means for recognizing. 2. A handwriting shorthand character input means is a tablet,
The handwriting shorthand character recognition device according to claim 1, further comprising an input pen for drawing a character on a drawing surface of the tablet. 3. A handwritten shorthand character input means for outputting coordinates of a writing point of an input pen on a tablet and a status signal indicating whether or not the input pen is in contact with the tablet. The handwritten shorthand character recognition device according to claim 2.