JPH07146918A - Handwritten character recognizing device - Google Patents

Abstract

PURPOSE:To facilitate the input of KANJI (Chinese character) by providing a stroke dictionary which contains all characters to be recognized while the characters are decomposed into strokes, collating input strokes with the stored strokes, and extracting characters which includes the input strokes as candidates of recognition. CONSTITUTION:Once a user selects one of recognized character candidates displayed on a display means 102 by using a candidate selecting means 110, the selected character is displayed as a recognition result. At the same time, the input order of strokes stored in a storage means 103 is compared with the input order in a writing stroke order dictionary and then the correct input order is displayed. The total number of strokes stored in the storage means 103, i.e., the number of strokes is compared with the number of strokes of the corresponding character stored in the storage means 103. When they do not match each other, a message showing that the number of strokes is wrong is displayed at a display means 102 and then the correct number of strokes is displayed. Consequently, the intended character can easily be obtained.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character processing apparatus having a function of inputting a character using a pen or the like and recognizing the character as a predetermined character, which can be input without the user knowing the stroke or stroke number. , And at any point during the input,
The present invention relates to a character input device capable of displaying candidate characters intended for input.

[0002]

2. Description of the Related Art Character recognition methods can be broadly classified into two types. One is offline character recognition, in which characters written on manuscript paper in advance are captured by a scanner etc. for recognition processing, and the other is offline characters in which characters are written with a pen on the tablet and recognition processing is performed in real time. Recognition.

As a recognition method, in offline recognition, a character is regarded as a set of light and shade (or black and white) points, and usually does not have information on strokes (strokes) or strokes. On the other hand, in the online character recognition, in general, in addition to the character shape information of the character, there is information on the stroke (stroke) and the number of strokes.

[0004]

In the online character recognition, the number of matching characters can be narrowed down because of the information on the strokes and strokes, which is not offline, and the processing speed can be improved and the recognition rate can be improved. I was able to plan. However, these features, which are not in offline recognition, can be clearly extracted when the input characters are written carefully, but there are many unstable elements.

That is, in a pen input device installed in a word processor or a personal computer, the subject of input is a general user, so for example, in the case of a slightly complicated kanji, there is no knowledge of the exact kanji shape, The stroke order of Kanji, such as strokes, is often ambiguous. Regarding the number of strokes, it is often the case that some Kanji characters are continued or omitted. It is rather rare. For these deformations, methods for predicting possible deformations have been proposed, but in all cases, since information on at least stroke order and stroke count is used, they are weak against deformations.

According to the present invention, when writing a character, even if the user does not care about the writing order, and even if he / she remembers the Chinese character incorrectly, he / she can input the knowledge and information that the user knows about the Chinese character. By doing so, it is possible to provide a character input device in which an intended character can be obtained by making maximum use of these and accurate knowledge (stroke order, character shape) can also be obtained.

[0007]

In order to solve the above problems, according to the present invention, a handwritten character input means for inputting strokes of handwritten characters one by one, and the input stroke (input stroke). Handwritten characters including storage means for storing, character recognition means for extracting recognition candidates for input characters based on the input stroke, and display means for displaying the recognition candidate characters extracted by the character recognition means. In the recognition device, the character recognition means, for all characters to be recognized, has a stroke dictionary that decomposes and stores the characters into strokes, and collates the input stroke with the strokes stored in the stroke dictionary, A character including at least the input stroke can be extracted as a recognition candidate.

[0008]

When the user starts inputting characters with a pen or the like, each time a stroke is input, it is collated with the stroke dictionary, and character candidates including the strokes input so far are displayed on the display device. When the input order of the characters is incorrect, the display device displays that the stroke order is incorrect, and the correct stroke order is displayed. If the characters are not correctly divided into strokes and input, the display device displays that the number of strokes is incorrect, and the correct number of strokes and how to divide the strokes are displayed.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the handwritten character input device of the present invention will be described below with reference to the drawings. A basic block diagram of the present invention is shown in FIG. The coordinate input means 101 is a tablet for inputting writing data of handwritten characters with an input pen or the like online.

The input handwriting data is stored in the storage means 103 in the order of input every time one stroke is input. The character recognition unit 104 refers to the stroke dictionary stored in the storage unit 103, refers to the stroke dictionary 105, and converts the stroke into one or a plurality of characters including the strokes. Display using 109. When the user uses the candidate selection means 110 to select one of the recognized character candidates displayed on the display means 102, the selected character is displayed on the display means 102 as a recognition result. At the same time, the stroke input order stored in the storage means 103 is compared with the input order of the corresponding characters in the stroke order dictionary. If they do not match, the display means 102 indicates that the stroke order is incorrect. Is displayed, and then a display showing the correct input order is executed. Subsequently, the total number of strokes, that is, the stroke number, stored in the storage unit 103 is compared with the stroke number of the corresponding character in the stroke number dictionary 107. If they do not match, the display unit 102 first indicates that the stroke number is incorrect. A message to that effect is displayed, and then a display indicating the correct number of strokes is executed.

In the above operation, the coordinate input means 101 is used to point out the error in the stroke order and display the correct stroke order, and to point out the error in the stroke number and display the correct stroke number. It has a switching means for selecting whether to perform or not.

When the stroke is sent, the character recognition means 104 collates with the stroke dictionary 105 to recognize the character. Reference numeral 108 is a control means for controlling each of the above means.

An external view of the present invention is shown in FIG. The apparatus of the present invention is roughly divided into a main body 201 that performs character input processing.
And a pen 202 for instructing coordinates, the main body 20
1 and the pen 202 are connected by a communication cord. The user holds the pen 202 and inputs / edits characters and figures by handwriting on the display-integrated tablet 203 of the main body 201. The pen 202 includes a main body and a power switch 204 for the pen. The main body 201 is equipped with an external storage device 205 such as an IC card or a floppy disk. Also, by equipping the pen side with the power supply of the main body,
It prevents the power from being accidentally turned off during input / edit operations, and also makes the body shape smooth and easy to carry. Here, the main body 201 and the pen 202 are
You may communicate by radio | wireless. In addition to the above, an execution key 206 for instructing execution of a process, a cancel key 207 for instructing cancellation of the process, and a direction key 208 used for moving the cursor and scrolling the screen are provided. The execute key 206, the cancel key 207, and the direction key 208 improve the operability of the user in that they are more intuitive than a soft key on the display screen for a beginner user and can always be used. You can Further, a selection switch 209 for selecting whether to display a message regarding the stroke order and the number of strokes of the present invention is provided.
The user presets this device before use.

Display examples of the input screen of the present invention are shown in FIGS. The user operates the character input box 302 on the screen 301 of FIG.
Input handwritten characters one stroke at a time. Figure 3 (a)
In the above example, the user has input the first strokes 303 and 304 of the Chinese character “not yet”. Screen 3
In the recognition candidate display portion 320 at the lower right of 01, character candidates including the strokes 303 and 304 are displayed. The user can further perform another two strokes (30
5, 306) If you input a lot, strokes 303 to 30
Character candidates including 6 are displayed. The user is shown in FIG.
As shown in, when the last stroke 307 of “not yet” is input, the recognition candidate display section 320 is as shown in the figure.
If 3 is selected, it is added to the text as 330. However, the user may, for example, 321 in the state of FIG. 3A and 322 in the state of FIG. 3B.
If you select, you can still get the character "un". In this way, the target character can be obtained by selecting the recognition candidate displayed in the recognition candidate display unit 320 without inputting all the strokes of the character. The above input and selection work correctly even if the user inputs the stroke order or the stroke count incorrectly.

Further, in the state of FIGS. 3 (a) and 3 (b),
As described above, when the character “not yet” is selected, the system is displayed as shown in the box 302 in FIG. 4 (a).
It also has the function of making "unfinished" characters. Alternatively, the completed character is not a handwritten image, and is shown in FIG.
The unprinted type character written in 330 may be used. Figure 4
(B) shows an operation in the case where the user incorrectly inputs the stroke order of characters. In FIG. 4B, the user makes a mistake in the writing order of “not yet” input,
5 shows an example of a screen display when inputting 5, 304, 306, 307 and selecting 323. As mentioned above, when 323 is selected, it collates with the stroke order dictionary,
Since it has determined that it is wrong, the system
The message "The stroke order is different. The correct stroke order is ↓" such as 1 is displayed, and at the same time, the display indicating the correct stroke order as indicated by 402 is executed.

FIG. 5 shows a state in which the user erroneously inputs the number of strokes of a character. Wrong strokes means that the kanji "power" should be written in two strokes, but the stroke of the first stroke is divided into two strokes, making it a total of three strokes, as shown in FIG. That is the case. As will be described later, the stroke dictionary 105 has a dictionary content that can be dealt with even when the strokes are divided as described above. In addition to 503, the character “power” 504 is also displayed as a recognition candidate. Here "power" 50
If you select 4, a message 50 warning you that the number of strokes is incorrect.
1 is displayed and the correct stroke information and stroke number information 502 are displayed.
Is displayed.

Next, the control means 10 will be described with reference to FIGS.
The processing flow of No. 8 will be described. In the following description, FIG.
And reference is made to the elements of the invention in FIG. In FIG. 6, the user uses the pen 202 to input strokes one by one (601). The input stroke data is additionally stored in the storage means 103 (602). Storage means 10
The stroke data stored in No. 3 is matched with the stroke dictionary 106 (a matching method will be described later) (603), and the matched one is determined to be a candidate for recognition up to that point (604). , The candidate is displayed on the candidate display section 320 of the display means 102 (60
5). If the user writes the next stroke at this point, the process returns to the one-stroke input of 601 and the above-described operation is repeated, but the user selects one of the candidates displayed on the candidate display unit 320. If (606),
The selected character is displayed in the text display area 350 of the display means 102 (607). If it is not a candidate selection,
Wait for next stroke input. Between 602-606,
The control means 108 senses whether or not there is a stroke input at a constant interval, and if there is a stroke input, the processing currently in progress is interrupted and the processing of 601 is started. By doing so, even if the user inputs strokes one after another, it is not necessary to display candidates each time, and the processing speed can be increased.

Next, a "stroke count completion flag" (for the structure of the stroke dictionary 105, corresponding to the selected character,
It is determined whether or not 1 (described later in FIG. 9) is 1 (61
4) If it is 1, the processes of 608 to 612 are performed, and if it is not 1, the process is terminated. The processing of 608 to 612 will be described below. "If the stroke count completion flag is 1,
The stroke information stored in the storage unit 103 is matched with the stroke order in the stroke order dictionary 106 corresponding to the selected character (a matching method will be described later) (608), and it is determined whether or not there is a match. Do (6
09). If no match is found in this matching, a message notifying that effect is displayed in the message display area 401 of the display means 102, and the correct stroke order information is displayed in 402 as described in FIG. 4 (61).
0). In addition, the stroke information stored in the storage unit 103 is used as the stroke number dictionary 107 corresponding to the selected character.
The number of strokes in the image is matched (a matching method will be described later) (611), and it is determined whether or not there is a match (612). If no match is found in this matching, a message notifying that effect is displayed on the display means 1
No. 02 is displayed in the message display area 401, and information on the correct number of strokes is displayed in 402 as described in FIG. 4 (613).

FIG. 7 shows another embodiment of the flow of control contents of the control means 108. The flow shown in FIG. 7 is similar to the control content of the control means 108 shown in FIG. 6, and only the differences will be described here. In FIG. 7, 7
The judgment indicated by 01 is added. At 602, when the stroke input by the user is added to the storage unit 103, the operation of accumulating stroke data is repeated until the user inputs the execution key 206 without performing matching with the stroke dictionary as described in FIG. . When it is determined that the user has input the execution key 206 (701), matching with the stroke dictionary (603) is performed in the order shown in FIG. By performing such an operation, the processing after 603 is not performed after the stroke input, so that the processing speed can be increased. in this case,
The control unit 108 does not need to sense whether or not a stroke has been input.

FIG. 8 is a diagram for explaining the contents and structure of data stored in the storage means 103, and FIG. 9 is a stroke dictionary 10.
5 is a diagram for explaining the structure of the dictionary of FIG. 5, and FIG. 10 is an embodiment showing a method of matching the stroke information stored in the storage means 103 with the stroke dictionary 105.

In FIG. 8, the data 802 is stored in the storage means 10 when the user inputs all strokes "one, one, |, /, \" in order to input the Chinese character "un".
3 shows the contents of 3. In the stroke time series data portion 803 of the data 802, all of the above strokes are
They are recorded in the order they were entered. In step 602 of adding to the stroke data (see FIGS. 6 and 7), in addition to the operation of recording the strokes in the input order as described above, the input strokes are set to a predetermined basic stroke (for example, in FIG. 8). Sort by 801),
In addition, the number of appearances of the stroke is indicated by a numerical value (804).

FIG. 9 shows an embodiment of the dictionary structure of the stroke dictionary 105. In the vertical direction indicated by 901, characters that are considered to appear are arranged. In the horizontal direction 902, types of predetermined basic strokes are arranged. 8 described in FIG.
The order of the basic strokes arranged in 01 is the same as the order of the basic strokes arranged in 902.

The stroke information table 903 for Chinese characters “not yet” shows information about how many strokes are included in each basic stroke 902.

[0024] For example, in the kanji "Mi", "1" is 2
That is, one “|”, one “/”, and one “\”.

FIG. 10 is a diagram for explaining in detail the procedure of matching 603 with the stroke dictionary in FIG. Figure 10
Will be described with reference to the specific examples shown in FIGS. A part of the dictionary structure of the stroke dictionary 105 in FIG. 9 has a column 904 indicating a candidate for a recognized character corresponding to the input stroke, and the symbol corresponding to the value of this column is MARK. . In addition, there is a column “stroke count completion flag” 912 that indicates whether the input stroke matches all stroke counts of the character. In the data structure of the stroke dictionary 105, the amounts other than MARK and the stroke number completion flag are values given in advance by the system.
K and the stroke number completion flag are as shown in FIG.
The amount written by the process.

In FIG. 10, first, the MARK value and the stroke number completion flag are all initialized to 0 (10
01). Next, information obtained by sorting the stroke information that has been input up to this point of time (“unknown stroke information 80
2) is taken out (1002). The information to be matched with the information is then extracted one by one. That is, the stroke data for one character 901 in FIG. 9 is extracted. Also,
The stroke number completion flag of the character is set to 1 (1003). In this case, it is assumed that the "un" stroke information 903 is extracted. Next, the input strokes extracted in steps 1002 and 1003 and the corresponding strokes of the reference stroke pattern are compared one by one. Now step 1002
The stroke number of the stroke with the input information extracted in step I
S, the number of strokes corresponding to the reference information extracted in step 1003 is set to SS (1004). In this case, IS = 2 because the value corresponding to “one” in the stroke information of 804 is IS. Further, in the stroke information table 903, the value corresponding to “one” is SS, so SS = 2. If SS = 0, the character just fetched cannot be a character representing the input stroke, so that the character is moved to another stroke without performing the matching operation for the stroke (1005). IS ≦ SS
If it is (1006), since this stroke of the input stroke is included in the stroke of the reference character, there is a possibility that it is a reference character, so a mark indicating that this stroke has been matched, that is, MARK = 1 is set. (1
007). In this case, IS = SS = 2, so 90
As shown in 6, the value in the MARK column 904 is set to 1. If IS> SS, the input stroke has no possibility of being the reference character currently set, so MARK
= 0 is set. Further, the stroke number completion flag = 0 is set (1008). Then the step of new reference character is entered. Further, when IS = SS, that is, when the input stroke matches all strokes of the matched character (1011), the stroke number completion flag is not set to 0. If IS <SS, the stroke number completion flag = 0 is set (1013). Next, it is determined whether there is a stroke that has not been compared (1009), and if there is a stroke, the process moves to the comparison of a new reference character, and if there is not, the reference character is changed to the next one. In this case, in addition to "one", "|"
Since "/", etc. remain, the above operation is performed for all of them. When all strokes are finished, it is judged whether or not there are still reference characters (1010),
If it remains, the process proceeds to the step of updating the reference character 1003. This means that all reference characters shown in 901 of FIG. 9 are performed. When all the reference characters are also finished, the step of decision 604 of the recognition candidate of FIG. 6 is entered.
This means that all the characters for which 1 is set in the MARK 904 for each character 901 in FIG. 9 are picked up. In the case of the present example, all three are “not yet”, “end” and “coming” 803.
Since these strokes are included, all three are set to MARK = 1 as shown in 904. If this is possible, the next step is to enter the candidate display 605 shown in FIG.

Reference numeral 806 in FIG. 8 shows time-series data of strokes when the first "not yet" three strokes are input. Since this data is included in the complete stroke data of "not yet", if 806 is matched with the stroke dictionary, MARK = 1 is set for "not yet", "end" and "coming" as in the case of 803. In this case, other characters are also candidates (for example, "shi", "soil", "work", etc.).

In the case of the stroke data 807,
This is an example of the case where the stroke order of the "unwritten" characters is incorrect. In this case as well, the time-series data 807 is sorted for each basic stroke, so the sorted stroke information 80 is sorted.
8 becomes exactly the same as the stroke information of 804,
The state of matching with the stroke dictionary is exactly the same as in 803. Step 6 of the flowchart shown in FIG.
As shown in 08, if the "un" character is selected,
Matching with the stroke order dictionary 106 is performed, but in this case, the stroke order does not match, so a message indicating that the stroke order is incorrect and the correct stroke order are displayed (6
10).

Further, 809 in FIG. 8 is stroke time series information of "force", but in the example shown here, the stroke number is erroneously input. That is, the correct stroke is "F"
Even though it is "|", write the first stroke on 2 strokes, and then "-"
In this example, "|" Stroke dictionary 105
Forms a dictionary with stroke patterns in which the number of strokes may be incorrect. For example, reference numeral 907 in FIG. 9 is an example of a correct stroke pattern of "force", and 908 is an example of a stroke pattern in which "force" is written in three strokes as described above. Even if a stroke of “force” as described above is input, it matches with the stroke pattern 908, and “force” is also displayed in the candidate display, so that the user can select “force”.
In this case, the input stroke is equal to 3 with respect to the total number of strokes 3 of the stroke pattern 908, so that the content of the stroke number completion flag 912 is set to 1. In this case, when the stroke number dictionary matching process 611 is performed, it is found that the stroke number is incorrect, and therefore a message indicating that the stroke number is incorrect and the correct stroke number are displayed (613).

FIG. 11 shows an embodiment of the stored contents of the stroke number dictionary 107 (see FIG. 1). The stroke number dictionary 107 is referred to only when the stroke number completion flag = 1 in the processing flow of FIG. 6, that is, when the user writes the total stroke number of the character. For information indicating the stroke number of the character selected by the user, by referring to the stroke number dictionary pointer in the stroke dictionary of FIG. 9, it is possible to know which part of the stroke number dictionary 107 should be referred to. As already mentioned, "power"
When the character is written in three strokes such as "-", "|", and "|", the stroke number completion flag in the stroke dictionary 105 is 1, so the stroke number dictionary 107 is referred to. In this case, the stroke number dictionary pointer points to 1101 in FIG. The number of input strokes is 3 with reference to 812 in FIG. 8, and the number of strokes of “force” is 2 with reference to 1102, which means no matching. Therefore, the result of 612 in the processing flow of FIG.
That is, "message display, stroke number information display" (613). The message and data for displaying the number of strokes have image data (1103) for each character,
This should be displayed.

FIG. 12 shows a stroke order dictionary 106 (see FIG. 1).
3 shows an example of the stored contents of the above. This dictionary is also referred to by the stroke number completion flag = in the processing flow of FIG.
It is 1 time.

As for the information indicating the stroke order of the character selected by the user, by referring to the stroke order dictionary pointer in the stroke dictionary of FIG. 9, it is possible to know which part of the stroke order dictionary 106 should be referred to. As already mentioned, when you write the letters "not" in the order "1, |, one, /, \",
Since the stroke count completion flag is 1, the stroke order dictionary 10
6 references are made. When matching is performed with the correct stroke order 1201 for the character “un” in FIG. 12,
The result of 609 in the processing flow of FIG. 6 is the result of “no matching”, which results in “message display, correct writing order display” (610). The message and the data for the correct writing pure display are held as image data (1202) for each character, and this may be displayed.

[0033]

As described above, according to the present invention, when a Chinese character or the like is input using a pen or the like, even if not all the strokes of the Chinese character are input, at the time of inputting a part of the stroke, Kanji candidates including the strokes are displayed, so you can easily input kanji.

Also, since the stroke order is completely free, you can enter the Chinese characters in any order, and even if you only remember the Chinese characters, you will be able to enter the candidates if you write only the part you know. It is convenient.

Furthermore, if the wrong kanji stroke or stroke number is entered at the time when the kanji character that appears as a candidate is selected in such a manner, the correct stroke order and stroke number information will be displayed on the screen. There is a user learning function, and the more this function is used, the more the user can learn the correct stroke order of kanji and the kanji itself, which is useful.

[Brief description of drawings]

FIG. 1 is a basic block configuration diagram of an embodiment of the present invention.

FIG. 2 is an external view of a character input device of the present invention.

FIG. 3 is an explanatory diagram showing a screen display example according to an embodiment of the present invention.

FIG. 4 is an explanatory diagram showing a screen display example according to an embodiment of the present invention.

FIG. 5 is an explanatory diagram showing a screen display example according to an embodiment of the present invention.

FIG. 6 is a flowchart showing the flow of processing according to an embodiment of the present invention.

FIG. 7 is a flowchart showing the flow of processing of an embodiment of the present invention.

FIG. 8 is an explanatory diagram showing the structure of time-series data of input strokes of the present invention.

FIG. 9 is an explanatory diagram showing the structure of a stroke dictionary according to the present invention.

FIG. 10 is a flowchart showing the flow of processing according to an embodiment of the present invention.

FIG. 11 is an explanatory diagram showing the structure of a stroke number dictionary of the present invention.

FIG. 12 is an explanatory diagram showing the structure of a stroke order dictionary of the present invention.

[Explanation of symbols]

 101 ... Coordinate input means 102 ... Display means 103 ... Storage means 104 ... Character recognition means 105 ... Stroke dictionary 106 ... Stroke order dictionary 107 ... Stroke number dictionary 108 ... Control Means 109 ... Candidate display means 110 ... Candidate selection means

 ─────────────────────────────────────────────────── ─── Continued Front Page (72) Hiroyuki Kumai, Inventor Hiroyuki Kumai, 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Hitachi, Ltd. Microelectronics Equipment Development Laboratory (72) Inventor Yasumasa Matsuda Totsuka-ku, Yokohama-shi, Kanagawa 292 Yoshidacho, Hitachi, Ltd. Microelectronics equipment development laboratory

Claims (8)

[Claims] 1. A handwritten character input means for inputting strokes of handwritten characters one by one, and the input stroke (input stroke).
Handwriting including storage means for storing, a character recognition means for extracting a recognition candidate of an input character based on the input stroke, and a display means for displaying the recognition candidate character extracted by the character recognition means. In the character recognition device, the character recognition means has a stroke dictionary that decomposes and stores the characters into strokes for all the characters to be recognized,
A handwritten character recognition device, characterized in that the input strokes are collated with the strokes stored in the stroke dictionary, and characters including at least the input strokes are extracted as recognition candidates. 2. The handwritten character recognition apparatus according to claim 1, wherein the stroke dictionary does not have stroke input order information. 3. The selection means for selecting one of the recognition candidate characters displayed on the display means according to claim 1, wherein the storage means also stores the order of the input strokes. The character recognition means also has a stroke order dictionary in which stroke stroke order information for each character is recorded, and stroke stroke stroke order information for the character selected by the selection means is acquired from the stroke order dictionary. A control means for collating the stroke order information with the order of the input strokes stored in the storage means, and displaying a message indicating that the stroke order is incorrect on the display device if they do not match. A handwritten character recognition device, comprising: 4. The handwritten character recognition device according to claim 3, wherein the control device displays the stroke order information as a message in addition to the message if they do not match in the collation. 5. The stroke dictionary according to claim 1, wherein one character is stored in association with a set of strokes decomposed by a plurality of disassembling methods. A handwritten character recognition device characterized by using at least one of a method of collectively forming a single stroke and a method of decomposing one stroke into a plurality of strokes. 6. The selection device according to claim 1, further comprising a selection unit for selecting one of the recognition candidate characters displayed on the display unit, wherein the storage unit also stores the number of strokes. The recognition means also has a stroke number dictionary recording stroke number information for each character, acquires stroke number information for the character selected by the selection means from the stroke number dictionary, and stores the stroke number information in the storage means. The handwritten character recognition device, characterized in that the input device is provided with a control means for collating the input stroke count of the input character and displaying a message indicating that the stroke count is incorrect if the stroke count does not match. 7. The control device according to claim 6, wherein the control device displays, as a message, the stroke number information recorded in the stroke number dictionary in addition to the message, if they do not match. Handwritten character recognition device. 8. The handwritten character recognition apparatus according to claim 3, 4, 6 or 7, further comprising means for designating a mode in which the message is displayed and a mode in which the message is not displayed.