JPH07182460A - Learning registering method/device for handwritten character recognition device - Google Patents

Abstract

PURPOSE:To prevent increase of the size of a learning dictionary by recognizing characters of the holograph data and rejecting the learning registration as long as the characters selected out of the candidate characters outputted in the prescribed priority are set in a prescribed sequence. CONSTITUTION:The holograph data on the characters inputted by handwriting through a character input part 101 are sent to a character recognition part 102 end a character registering part 105. The part 102 sends successively the received holograph data to a candidate character display part 103 and a character selecting part 104 in the order of data recognized closer to the stroke data stored in a dictionary against the corresponding characters. The part 103 displays all of candidate characters. Then an input operator decides his desired characters out of the candidate characters and then requests the learning registration against the holograph data. Thus the part 104 rejects reception of the selected characters as long as they are set in a fixed sequence in the displayed candidate characters end does not register the characters in a learning dictionary 106.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a learning handwritten character recognition device, and more particularly to a learning registration method and device.

[0002]

2. Description of the Related Art In a handwritten character recognition apparatus, an input character is selected and designated as a character intended for input. That is, as a method for confirming, as described in "Japanese Patent Laid-Open No. 2-207
No. 387 ”and the like, there is a method of displaying a plurality of candidate characters obtained as a result of character recognition and selecting an intended character from among them with a pen. Similarly, this method can be used when a character is registered for learning. Further, as a method of designating a candidate character when the character to be learned and registered does not exist, see Japanese Patent Laid-Open No.
-103787 "and the like.

Learning registration of a conventional handwritten character recognition apparatus will be described below with reference to the drawings. FIG. 9 is a block diagram mainly showing a learning registration device of a conventional handwritten character recognition device. In FIG. 9, 901 is a character input unit, 902 is a character recognition unit, 903 is a candidate character display unit, 904 is a registered character designation unit, 905 is a character registration unit, and 906 is a learning character. It is a dictionary section.

In addition to the above, not only the basic dictionary section in which the data of the lines and the points constituting them are stored in association with each character that is frequently input in advance, but also the keyboard and the floppy disk storage section are stored. , Which has the necessary components such as the printing unit to function as a handwritten character input device,
Further, not only each component of the learning registration device but also the learning registration device itself is integrally incorporated in the handwritten character input device. However, since these are self-explanatory, the illustration and description of each component, and the integrated configuration A description of specific contents is omitted. And this is the same also in the later examples.

The operation of each part of the conventional handwritten character recognition device will be described below. First, in the character input unit 901, handwriting data of handwritten characters is input by the input person using a pen and a tablet. The input handwriting data is passed to the character recognition unit 902 and the character registration unit 905, and the character recognition unit 902 performs character recognition. As for the pen and tablet, see, for example, "Shigeru Watanabe et al., Microcomputer Handbook, Vol. II, Vol.
This is a well-known technique described in “Chapter” and the like. For information on handwriting recognition technology, see “(Incorporated Association) The Institute of Electronics, Communication and Communication, Online Handwritten Kanji / Hiragana Recognition ~
Tetsuo Tomimoto, Kimiko Shima, Yutaka Ohta; P. P. published on October 15, 1982. 49-56 ”and the like, and description thereof will be omitted because these are so-called well-known techniques.

As a result of character recognition, candidate characters of recognition results for input handwriting data are listed. When there are a plurality of candidate characters, data in the order of high recognition, that is, handwritten lines registered in association with the dictionary (strokes, strokes, strokes, etc., which are figures that compose this, lines, points, etc.). Is passed to the candidate character display unit 903 in order from the character recognized as closest to the handwriting data input by handwriting. The candidate character display unit 903 displays all the characters selected as candidates by the character recognition unit 902.

Next, in the actual input of handwritten characters,
At this stage, as with a keyboard-type word processor, the character the user intended to input is selected from the displayed candidate characters, and then the character is inserted at the specified position in the text being created. Mostly displayed. Further, similarly, prior to learning registration, the inputter performs work such as referring to the display of candidate characters to determine whether or not to register handwriting data. However, since these are so-called obvious matters, detailed description of this portion will be omitted. And this is the same also in the later examples.

The operation of the present invention will be described below. The registered character designating unit 904 causes the inputter to designate a specific character for learning and registering the handwritten data.
The designated method is to select the intended one from those displayed in the candidate character display portion 903, to use the handwritten characters again in the regular writing, and to use the correctly recognized characters based on this, and to input the kana from the keyboard. To convert to Kanji, and also enter the Kanji code. The registered character designation unit 904, when the registered character is designated by the input person,
This is notified to the character registration unit 905. Character registration unit 905
Is the registered character received from the registered character designation unit 904,
The handwritten handwriting data input by the character input unit 901 or the same handwriting data separately input for learning is associated with each other and registered as new data in the learning dictionary unit 906. The character data registered in the learning dictionary unit 906 is
The character recognition unit 902 is used for subsequent character recognition. Therefore, when the same handwriting is written, the character registered here appears as a recognition candidate.

[0009]

In the [0005] However, as described above methods, the handwriting input person tends to learn many characters by just selecting or designated for speeding the handwriting input. Moreover, even if a correct recognition result is obtained and the character intended for input is displayed as a candidate character, the dictionary is often registered. Therefore, the data registered in the learning dictionary unit may increase more than necessary.
This not only increases the amount of storage resources required by the learning dictionary unit, and thus increases the search time,
Since handwriting data that are similar to each other are registered, as a result, they may interfere with each other, impairing the learning effect. Specifically, even if the same character is intended to be input, the handwriting data input by handwriting are different from each other in the strict sense, and therefore, it is necessary to check each stroke or check the positional relationship between them. For this reason, a character that is already registered may be displayed as a candidate character, or conversely, a stroke of the registered character may be detected when another character is input by handwriting, which may cause misrecognition. . In addition, "X" and "x" and "Tachi" and "shi" are clearly different symbols, characters, and kanji, but depending on the input person, they may be displayed as figures. In some cases, the characters may be entered and registered in a similar manner, and both kanji may be displayed in subsequent handwriting input, which may even hinder the effect of registration by learning. Furthermore, the stroke as a part of the handwriting data that constitutes the registered character is erroneously recognized when the character recognition unit narrows down the candidate characters based on each stroke input as the first step of the recognition work. May occur. For example, in the case of the above example, it may occur with kanji such as "Shu" and "Mori". However, since these are so-called well-known techniques known from, for example, Japanese Patent Application No. 5-309334 “Handwriting symbol recognition method and apparatus having learning ability”, further description will be omitted. On the other hand, selecting the character displayed as the candidate character in the first or second place is the character originally intended for the input by selecting from the candidate characters displayed in recent years. It is extremely easy and quick due to the improvement of the technology when selecting the input character as the intended character. As a specific example of such a technique, a selection button is provided adjacent to the lower part of the display position of the first and second characters, or a transparent button is provided on the upper surface of the display position. The same effect as pressing can be achieved by just touching the pen.

In view of the above problems, the present invention has been made for the purpose of providing a learning registration method and apparatus for handwritten character recognition which prevents an increase in the size of the learning dictionary by improving the learning registration of characters. is there.

[0011]

In order to solve the above problems, in the invention of claim 1, a character input step of inputting handwritten characters, and handwriting data input by handwriting are collated with data in a dictionary section. Character recognition step for outputting the candidate characters in a predetermined priority after performing character recognition, a candidate character display step for displaying the candidate characters output by the character recognition step in accordance with the order, and a candidate character display step for displaying A new character is created by associating the character selected by the registered character selection step with the registered character selection step in which the input person selects a character for which learning registration is requested from among the candidate characters selected, and the handwriting data input by handwriting. A learning registration method of a handwritten character recognition device, comprising: a character registration step of registering in a dictionary unit as recognition data, When the characters selected in the recorded character selection step are candidate characters up to a certain rank among the characters displayed in the candidate character display step, there is a registration refusal step of rejecting registration in the character registration step. It is characterized by that.

According to another aspect of the present invention, a character input step of inputting a handwritten character and a character recognition for performing a character recognition using handwriting data input by handwriting and then outputting candidate characters in a predetermined priority order. Step, a candidate character display step that displays the candidate characters output by the character recognition step according to their rank, a registered character specification step that allows the input person to specify the character to be learned and registered separately, and handwriting data that is input by handwriting is registered. In the learning registration method of the handwritten character recognition device, which has a character registration step of newly registering in the dictionary section as data for character recognition in association with the character specified by the character specification step, the specified character is specified in the registered character specification step. If the selected character is a candidate character up to a certain rank among the characters displayed in the candidate character display step, the character registration is performed. It is characterized by having a registration rejection step of rejecting the registration in step.

In the inventions of claims 3 and 4,
The fixed order at which the registration refusing step rejects the registration,
It is characterized by being only the first place. According to the invention of claim 5, the character input unit for inputting handwritten characters, the dictionary unit, and handwriting data input by handwriting are collated with the data in the dictionary unit to output the data, and the character recognition is performed, and then a predetermined value is obtained. The input person learns from the character recognition unit that outputs candidate characters in priority order, the candidate character display unit that displays the candidate characters output by the character recognition unit according to the order, and the candidate characters displayed by the candidate character display unit. The registered character selection unit that selects the character that requires registration, and the handwriting data that is input by handwriting, is associated with the character selected by the registered character selection unit, and is registered as new character recognition data in the dictionary unit. In a learning registration device of a handwritten character recognition device having a copy part, the characters selected by the registered character selection part are candidate characters up to a certain rank among those displayed on the candidate character display part. It is characterized in that it sometimes has a registration rejection unit for rejecting the registration in the character registration unit.

According to the invention of claim 6, character recognition is performed by collating the handwriting input character input section for inputting handwritten characters, the dictionary section, and the handwriting data input by handwriting with the dictionary section, and then predetermined priority is given. A character recognition unit that outputs candidate characters in order, a candidate character display unit that displays the candidate characters output by the character recognition unit according to the order, a registered character specification unit that allows the inputter to specify the character that you want to register for learning, and handwriting In the learning registration device of the handwritten character recognition device, which has a character registration unit that associates the handwriting data input in (1) with the character designated by the registered character designation unit and newly registers it in the dictionary unit as data for character recognition, Registration rejecting registration in the character registration unit when the characters designated in the registered character designation unit are candidate characters up to a certain rank among those displayed in the candidate character display unit It is characterized by having a absolute unit.

In the inventions of claims 7 and 8,
The registration refusal unit is characterized in that the fixed order of refusal of registration is only the first rank.

[0016]

With the above structure, in the invention of claim 1,
In the registration refusal step, when the characters selected for registration by the inputter in the registered character selection step are candidate characters up to a certain rank among the characters displayed in the candidate character display step, the inputter's registration request Nevertheless, it refuses registration.

In the invention of claim 2, in the registration refusal step, the character designated as the registration target by the input person in the registered character designation step is displayed up to a certain rank among the characters displayed in the candidate character display step. If the character is a candidate character, the registration in the character registration step is rejected. Claim 3
Further, in the invention of claim 4, the fixed order in which the registration refusing step rejects the registration is only the first rank.

In the invention of claim 5, the registration refusal section is:
When the character selected as the registration target by the input character in the registered character selection unit is a candidate character up to a certain rank among the characters displayed in the candidate character display unit, the registration in the character registration unit is rejected. In the invention of claim 6, when the registration refusal unit is a candidate character up to a certain rank among the characters designated by the input character as a registration target by the input person in the candidate character display unit, , Refuses registration at the character registration unit.

In the inventions of claims 7 and 8,
The registration refusal department rejects registrations at a certain rank, which is the first rank.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A learning registration method and device for a handwritten character recognition device according to the present invention will be described below based on embodiments. (First Embodiment) FIG. 1 is a block diagram of the first embodiment of the present invention. In the figure, 101 is a character input unit, and 102
Is a character recognition unit, 103 is a candidate character display unit,
Reference numeral 104 is a character selection unit, 105 is a character registration unit, and 106 is a learning dictionary unit. FIG. 2 is an operation flow chart of the character selection unit 104 shown in FIG. FIG. 3 is a diagram showing an example of handwriting data of handwritten characters input to the learning registration device shown in FIG. 1 and a display example of a character recognition result. In this figure,
(A) is an example of handwriting data of handwritten characters input by the character input unit 101, and similarly (B) is a candidate character in the order in which the candidate character display unit 103 has a high degree of recognition for the handwriting data of (A). It is a display example at the time of displaying. 4 is shown in FIG.
FIG. 6 is an explanatory diagram showing a state in which the learning registration unit for handwritten character recognition shown in FIG. 6 registers a character requested to be registered in the learning dictionary unit 106, and reference numeral 106 in this figure is also a learning dictionary unit.

Next, the operation of the learning registration apparatus for handwritten character recognition in the present embodiment configured as described above will be described with reference to FIGS. 1 to 4. First, in the character input unit 101, a handwriting character is input by the input person using a pen and a tablet. The handwriting data of the input handwritten character is the character recognition unit 102 and the character registration unit 10.
5, and the character recognition unit 102 performs character recognition.
In this embodiment, the same technique as the conventional example is used for the handwriting character recognition technique itself. As a result of character recognition, candidate characters are included in the input handwriting data. When there are multiple candidate characters, the candidate characters are displayed in descending order of recognition degree, that is, from the one in which the input handwriting data is found to be closest to the stroke data stored in the dictionary in association with that character. Part 1
03 and the character selection unit 104. However, only the first-ranked candidate characters are not passed to the character selection unit 104 in order to prevent registration. The candidate character display unit 103 displays all the characters selected as candidates by the character recognition unit 102.
The character selection unit 104 causes the input person to select which of the displayed candidate characters the handwritten character, that is, the handwriting data, should be learned and registered.

The operation flow of the character selection unit 104 will be described below with reference to FIG.
Follow along. (201) First, the input person is made to select a character to be registered. (202) Next, it is checked whether or not this character is the first character among the candidate characters listed by the character recognition unit 102, just in case. (203) If it is not the 1st place (N in 202)
o) The character registration unit 105 is caused to register the selected character, and thereafter, the main registration operation ends. If it is in the first place (Yes in 202), this character is not selected and the registration is not performed.
After calling attention in, the operation of the main registration request ends without registration.

Next, regarding the operation and processing associated with registration, when the character registration unit 105 receives a registration request from the character selection unit 104, the handwriting data input by the character input unit 101 and the character selection unit are input. The character selected as the registration target in 104 is registered in the learning dictionary unit 106 as new data in association with the character. Then, this data registered in the learning dictionary unit 106 is used for subsequent character recognition in the character recognition unit 102. And when the same handwriting is input,
After confirmation of the identity with the registered handwriting data, the corresponding registered characters are displayed as the recognition candidates.

The learning operation of a handwritten character input to the character input unit 101 as shown in FIG. 3A will be specifically described below. The character recognition unit 102 receives the handwriting data as shown in FIG. 3A and performs character recognition. The candidate characters of the recognition result are passed to the candidate character display unit 103 in descending order of recognition rate. Candidate character display unit 103
The recognition result is displayed as shown in FIG. 3B in order to present the recognition result to the input person so as to select an intended character or to receive a learning instruction. The display is arranged in order from the character having the highest recognition rate in the columns (1) to (4). In the case of learning, the input person determines the character intended by himself from these candidate characters, and on this, makes the learning registration corresponding to the handwriting data. If the input person selects "Hori" in (3), the character selection unit 104 checks whether or not "Hori" is the first in the display order of candidate characters, just in case. In this case, the first
Since it is not in the rank, as shown in FIG.
Registration to 6 is done. Further, when the "fence" of (1) is selected, the "fence" is the first place, so the acceptance is rejected, and as shown in (b) of FIG. 4, the learning dictionary 106 is not registered. . (Second Embodiment) FIG. 5 is a block diagram of the second embodiment of the present invention. In FIG. 5, 501 is a character input unit,
2 is a character recognition unit, 503 is a candidate character display unit, 504 is a registered character designation unit, 505 is a character registration unit, and 506 is a learning dictionary unit. FIG. 6 is an operation flowchart of the registered character designating unit 504 of the learning registration device of FIG. FIG. 7 is an explanatory diagram of an example of handwriting data of handwritten characters input to the learning registration device shown in FIG. 5 and an example of candidate characters of the character recognition result. In this figure, (A) is the character input unit 5
It is an example of handwriting data of handwritten characters input in 01,
Similarly, (B) is a character recognition unit 50 for this handwriting data.
2 shows candidate characters in descending order of recognition degree. FIG. 8 is an explanatory diagram showing a state in which the learning registration unit shown in FIG. 5 registers the characters requested to be registered in the learning dictionary unit 505. Also in this figure, 506 is a learning dictionary unit.

Next, the operation of the learning registration device of the handwritten character recognition device of the present embodiment configured as described above will be described with reference to FIGS. 5 to 8. First, in the character input unit 501, a user inputs a handwritten character using a pen and a tablet. The input handwriting data of the handwritten character is the character recognition unit 502 and the character registration unit 505.
And the character recognition unit 502 performs character recognition. In this embodiment, the same technique as the conventional example is used for the recognition technique itself of handwritten characters. As a result of the character recognition, candidate characters are included in the input handwriting data. When there are a plurality of candidate characters, the candidate character display section 5 is arranged in order from the character recognized as being closest to the input handwriting data.
Pass to 03. In addition, the first-ranked candidate character is also passed to the registered character designating unit 504. In the present embodiment, it is impossible to register the candidate character in association with the input handwriting data as it is because of the hardware. Therefore, the registered character designation unit 504 separately causes the input person to designate which character the handwritten data is to be learned and registered. Here, the same character designating method as in the conventional example is used. Hereinafter, the operation of the registered character designation unit 504 will be shown in the flowchart of FIG. (601) First, the input character is made to input the registered character. (602) Next, it is checked whether or not the registered character is the first character among the candidate characters listed by the character recognition unit 502. (603) If it is not the 1st place (N in 602)
o) The character registration unit 505 registers the designated character. As a result, the learning registration operation ends.

If it is the first candidate (Yes in 602,
Characters are not registered in s). In addition, this operation ends. Next, regarding the content of the registration operation, the character registration unit 505 specifies the handwriting data input in the character input unit 501 and the registration character specification unit 504 when a registration request is made from the registration character specification unit 504. The registered character is registered in the learning dictionary unit 506 as new data in association with the registered character.

Hereinafter, taking as an example a case where a handwritten character as shown in FIG.
This operation will be specifically described. The character recognition unit 502 receives the handwriting data shown in FIG. 7A and performs character recognition. The candidate characters of the recognition result are arranged in descending order of recognition rate, as shown in FIG. Then, the candidate character display section 5
These candidate characters are passed to 03, and at this time, the first candidate character is also passed to the registered character designating section. Now, the registered character designating unit 504 causes the input person to designate a character to be registered, and makes the learning registration of the character. Then, if the input person specifies “moat”, the registered character designating unit 504 checks whether or not this “moat” is the first candidate character. in this case,
Since "Hori" is not the first candidate, it is registered in the learning dictionary unit 506 in association with the handwriting data as shown in FIG. 8A. In addition, when the input person selects "grip",
Since "grip" is the first place, it is not registered in the learning dictionary unit 505 as shown in (b) of FIG. The character data registered in the learning dictionary unit 505 is used by the character recognition unit 502 for subsequent character recognition, and when the same handwriting data is input, the character registered here appears as a recognition candidate. Needless to say. (Third Embodiment) The third embodiment of the present invention will be described below.

The present embodiment is an improvement of the first and second embodiments described above, and is different only in the following points. (1) As the display order of candidate characters, not only the degree of coincidence between handwriting data input by handwriting and stroke data registered in advance in the dictionary unit in association with characters, but also the result of this degree of coincidence It also reflects the frequency of use of what was said. (2) The registration refusal unit rejects registration of candidate characters, which are candidates for character registration and are not ranked first in the display order but second.

Therefore, the description of the configuration and operation of each part will be omitted with reference to the configuration diagram, and the configuration, operation principle and effect of only the above two points will be described. (1) Regarding display order of candidate characters. Generally, the display order of candidate characters is calculated by collating the handwriting data input by handwriting with the strokes registered in the basic dictionary etc. for each character in correspondence with each other, and the matching degree of each character is quantified. The values are determined in descending order.

Specifically, for example, the difference in the length of the horizontal line or the vertical line is digitized into a numerical value from 1 to 0 by a non-linear function, and the presence or absence of the stroke itself is set to a numerical value of 1 or 0. By multiplying all of these numerical values with, the degree of coincidence for each character is given as a numerical value within the range of 0 to 1. In the first and second embodiments, the values are displayed in descending order. However, in this embodiment, if the digitized difference in the degree of coincidence is within 0.05, a flag indicating that the candidate character having a small degree of coincidence is frequently used as a result of long-term learning is added. The characters having a high degree of coincidence are displayed as the previous rank only when the flag is not attached. In addition, as a result of long-term learning, attach a flag to the characters that are frequently used, refer to the presence or absence of this when displaying.
Comparing the difference in the degree of coincidence is a well-known technique. Therefore, the description of the specific configuration for exerting those actions is omitted. (2) Regarding refusal to register the candidate character ranking up to 2nd place.

When a registration instruction is given, the display order of the candidate characters selected or designated under the instruction is referred to, and if it is at least two, it is rejected. Even if the registration is refused up to the second place, the selection of the character intended for input by the character input person is performed by touching the selection instruction part provided adjacent to the upper surface or the lower part of the display part of the character with a pen. Needless to say, it does not hinder the promptness in inputting handwritten characters because it is performed by such a simple operation. Moreover, in this case, since the selection buttons are provided only on the upper two candidate character display portions, the increase in the initial cost is slight. However, the provision of a button on the upper surface of the displayed candidate character that allows input of instructions by touching is a well-known technology that is also used in automatic cash dispensers at banks, etc. To do.

Although the present invention has been described above based on the embodiments, the present invention is not limited to the above embodiments, and the following things are of course included in the present invention. is there. (1) For the convenience of manufacturing, programming, etc., one constituent element (requirement) of the present invention may be plural, or conversely, a plurality of constituent elements (requirement) may be integrated, or these are appropriately combined. . (2) Modify the existing handwriting recognition device,
The operation of the present invention is performed by adding a device or a program for exhibiting the functions of the present invention. (3) Handwritten characters are not limited to "kana", "kanji", "alphabet", "arabic numeral", "onmon", "german", etc. Syllabary used in ","
Including symbols such as vertical line “│”, horizontal line “─”, and other symbols and gestures for pointing. (4) For any appropriate intention of the person who entered the handwritten character,
A function of stopping the exertion of the action according to the present invention by the selection operation of the person is added. (5) The learning dictionary unit is not installed separately, but is integrated into the basic dictionary unit that is incorporated as an essential item in the handwriting word processor. Alternatively, the memory resources that have been used as hardware are diverted. (6) In learning of handwriting data, not only is the entire input data registered as one character, but a function is also added in which it is decomposed into strokes and then learned for each stroke.

[0033]

As described above, according to the present invention, when a character is registered corresponding to the handwriting data input by the input person, the character is recognized by the character recognition means, the frequency of use, etc. When the candidate character having a high priority among the candidate characters displayed in consideration of, the registration in the dictionary unit is not performed. For this reason, it is possible to prevent the learning dictionary unit from unnecessarily increasing in size, maintain the quickness of the search, and eliminate erroneous recognition due to mutual interference of registered handwriting data.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a first embodiment of a learning registration device for handwritten character recognition according to the present invention.

FIG. 2 is an operation flow chart of a character selection unit of the first embodiment.

FIG. 3 is an explanatory diagram of a display example of input handwriting data of handwritten characters and candidate characters of a character recognition result in the first embodiment, and (A) is an example of input handwriting data of handwriting characters. Yes, (B) is an example in which candidate characters are displayed in descending order of recognition degree by the candidate character display unit.

FIG. 4 is a diagram showing a state in which a character requested to be registered is registered in a learning dictionary unit in the first embodiment.

FIG. 5 is a configuration diagram of a second embodiment of a learning registration device for handwritten character recognition according to the present invention.

FIG. 6 is an operation flow chart of a registered character designating unit 503 of the second embodiment.

FIG. 7 is an explanatory diagram of a display example of handwriting data of input handwritten characters and candidate characters of a character recognition result in the second embodiment, and (A) is an example of handwriting data of input handwriting characters. Yes, (B) shows the result of character recognition in the descending order of recognition degree.

FIG. 8 is a diagram showing a state in which a character requested to be registered is registered in a learning dictionary unit in the second embodiment.

FIG. 9 is a block diagram of a conventional learning registration device for handwritten character recognition.

[Explanation of symbols]

 101, 501, 901 Character input unit 102, 502, 902 Character recognition unit 103, 503, 903 Candidate character display unit 104 Registered character selection unit 504, 904 Registered character designation unit 105, 505, 905 Character registration unit 106, 506, 906 Learning dictionary section

Claims (8)

[Claims] 1. A character input step of inputting a handwritten character, and character recognition by performing handwriting recognition by collating handwriting data input by handwriting with data of a dictionary unit and then outputting candidate characters in a predetermined priority order. Step, and a candidate character display step that displays the candidate characters output by the character recognition step according to their rank, and a registered character that selects the character that the input person requests learning registration from among the candidate characters displayed by the candidate character display step A handwriting character recognizing device having a selecting step and a character registering step of newly associating the character selected by the registered character selecting step with the handwriting data input by handwriting and newly registering it in the dictionary unit as data for character recognition. In the learning registration method, the character selected in the registered character selection step is displayed in the candidate character display step. It has been when a candidate characters up to a certain order in the learning method for registering handwriting recognition apparatus characterized in that it has a registration rejection step of rejecting the registration in the character registration step. 2. A character input step of inputting handwritten characters, a character recognition step of performing character recognition using handwriting data input by handwriting, and then outputting candidate characters in a predetermined priority order, and a character recognition step. The candidate character display step of displaying the output candidate characters according to their rank, the registered character specification step of allowing the input person to specify the character to be learned and registered, and the registered character specification step of the handwriting data input by handwriting In a learning registration method of a handwritten character recognition device having a character registration step of newly registering in a dictionary unit as data for character recognition in association with a character, the character designated in the registered character designation step is the candidate character display. If it is a candidate character up to a certain rank among those displayed in step, reject the registration in the character registration step Learning how to register a handwritten character recognition device, characterized in that a registration rejection step that. 3. The learning registration method for a handwritten character recognition apparatus according to claim 1, wherein the fixed order of refusing registration in the registration refusing step is only the first rank. 4. The learning registration method for a handwritten character recognition device according to claim 2, wherein the fixed order of refusing registration in the registration refusing step is only the first rank. 5. A character input unit for inputting handwritten characters, a dictionary unit, and handwriting data input by handwriting is collated with data in the dictionary unit for character recognition, and then candidate characters are selected in a predetermined priority order. The character recognition unit that outputs, the candidate character display unit that displays the candidate characters output by the character recognition unit according to their rank, and the character that the inputter requests learning registration from among the candidate characters displayed by the candidate character display unit. Handwritten characters having a registered character selection unit to be selected and a character registration unit for newly registering in the dictionary unit as data for character recognition by associating the handwriting data input by handwriting with the character selected by the registered character selection unit In the learning registration device of the recognition device, when the character selected by the registered character selection unit is a candidate character up to a certain rank among the characters displayed on the candidate character display unit, Learning registration device of handwritten character recognition apparatus, characterized in that a registration rejection unit for rejecting the registration in shape registration unit. 6. A character input unit for inputting handwritten characters, a dictionary unit, and handwriting data input by handwriting is collated with the dictionary unit to perform character recognition and then output candidate characters in a predetermined priority order. A character recognition unit, a candidate character display unit that displays the candidate characters output by the character recognition unit according to their rank, a registered character specification unit that allows the input person to specify the character to be learned and registered separately, and handwriting data input by handwriting. In the learning registration device of the handwritten character recognition device, which has a character registration unit that associates with the character designated by the registered character designation unit and newly registers in the dictionary unit as data for character recognition, the registered character designation unit When the designated character is a candidate character up to a certain rank among the characters displayed in the candidate character display section, it must have a registration refusal section that rejects registration in the character registration section. Learning registration device of handwritten character recognition apparatus according to claim. 7. The learning registration device of the handwritten character recognition device according to claim 5, wherein the fixed order of refusal of registration by the registration refusal unit is only the first rank. 8. The learning registration device of the handwritten character recognition device according to claim 6, wherein the fixed order at which the registration refusal unit rejects the registration is only the first rank.