JPH06309490A - Character input device and its system - Google Patents

Abstract

PURPOSE:To reduce the number of user's operation steps necessary for character input and correction in a hand-written input device. CONSTITUTION:Written data inputted from a coodinate input device 101 are converted into a character code by a character recognizing means 104 and then converted into 'KANJI' (Chinese character) by a character conversion means 105 and the converted result is displayed. When the displayed character string is not a required one, a user instructs to correct the character concerned, so that the conversion result of combinations of all candidate recognition data stored in a storage means 103 and recognition candidates are simultaneously displayed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character input device such as a word processor for inputting characters, and more particularly to a device for further kana-kanji conversion of character recognition results, especially handwritten character recognition results.

[0002]

2. Description of the Related Art Recently, a character recognition device for converting printed characters, images of handwritten characters on paper read by a scanner, or stroke data directly written on a coordinate input device such as a tablet into character codes. Performance increased,
In addition to English characters and Katakana, it has become possible to recognize many characters such as Hiragana and Kanji. In particular, a handwriting input (pen input) method using a tablet has begun to be adopted as an input means replacing a conventional keyboard in a word processor, a personal computer or the like.

However, when a user creates a document,
Since not all desired kanji can be written, and some kanji have a large number of strokes and handwriting input is difficult, the device in which the character recognition means is the only character input means is inconvenient.

Therefore, a method is known in which kana-kanji conversion means is prepared and kana characters input by character recognition are converted into kanji characters so that desired characters can be obtained. In addition, at this time, a method is known in which conversion accuracy is improved by performing conversion in consideration of the kanji before and after the kana character.

By the way, in general, the accuracy of character recognition in pen input is about 90% and not 100%. Therefore, if the recognition result output by the character recognition means is directly converted into Kana-Kanji, the intended result may not be obtained. For example, when the user inputs "island" and intends to write "shima" but the recognition result is "kuma", the user can obtain "bear" by kana-kanji conversion,
Cannot enter "island". Also, the kana-kanji conversion accuracy is around 90%, not 100%. Therefore, the conventional device is designed to allow the user to input characters in the following steps.

(Step 1) Write characters on the tablet.

(Step 2) Character recognition is performed.

(Step 3) Confirm the correctness of the recognition result,
If it is incorrect, a correction instruction is given and the correct recognition result is selected from the displayed recognition candidates.

(Step 4) Kana-Kanji conversion is performed.

(Step 5) Confirm the correctness of the conversion result,
If it is incorrect, a correction instruction is given, and the correct conversion result is selected from the displayed conversion candidates.

[0011]

The above-mentioned conventional method is the first one.
There was a problem that there were many operation steps.

Secondly, if the result shown in (step 5) is different from the result desired by the user, whether the factor is a Kana-Kanji conversion error or not, a correct recognition result can be selected in (step 3). There was a problem that the user could not make a clear decision, probably because it was not there. That is, when the character "Kuma" is displayed as the conversion result,
If the user mistakenly thinks that this character is read as "shima", the user will search for "island" and search for conversion candidates. Conversely, if the character "Shima" is displayed as the conversion result, "shima" will be displayed. If the user misunderstands not to read, the user tries to return to (step 1) or (step 3), which results in an unnecessary operation.

Thirdly, since it is impossible to return from (step 5) to (step 3) in the conventional apparatus, if an error in the recognition result is found in (step 5),
There was a problem that it was necessary to start over from handwriting input.

An object of the present invention is to integrate (step 2) and (step 4), (step 3) and (step 5), respectively, so that a character input device and method can be used which allows natural character input with a small number of operation steps. To provide.

[0015]

In order to achieve the above object, in the character input device of the present invention, a storage means for storing recognition candidates and a character conversion means for performing kanji conversion for all combinations of the stored recognition candidates. Was set up.

[0016]

The writing data input from the coordinate input means is converted into a character code by the character recognition means, converted into kanji characters by the character conversion means, and the conversion result is displayed. When the displayed character string is not a desired character, the user gives a correction instruction to the character, thereby simultaneously displaying the result of conversion of all combinations of the recognition candidate data stored in the storage means and the recognition candidate.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the character input device of the present invention will be described below with reference to the drawings.

A basic block diagram of the system of one embodiment of the present invention is shown in FIG. The coordinate input means 101 may be a tablet that inputs writing data of handwritten characters online, or a scanner that inputs documents offline. The input handwriting data is converted into a character code by the character recognition means 104. The converted character string is the character conversion means 10
By 5, the kana part that can be converted into kanji is converted into kanji. The recognition result of the character recognition means 104 and the character conversion means 1
The recognition result of 05 is stored in the storage unit 103, and the display unit 1
02 is displayed. Reference numeral 106 is a CP that controls each of the above means.
Control means such as U.

An external view of an embodiment of the present invention is shown in FIG.
The apparatus of the present invention is roughly divided into a main body 201 for inputting characters and a stylus 202 for pointing coordinates. The user has a stylus 202, and the main body 201
Characters and figures are input by handwriting to the display-integrated tablet 209. When converting the input characters to Kana-Kanji, the conversion key 203 provided on the stylus 202 is
It has the same function as the conversion instruction button of the on-screen soft key, and can be used by the user or in some cases. The main body 201 includes a power switch 20 for the device.
4. An external storage device 208 such as a floppy disk / IC card is provided. In addition to this, an execution key 205 for instructing execution of a process and a cancel key 20 for instructing cancellation of a process
6. A direction key 207 used for cursor movement and screen scrolling is provided. Enter key 205 / Cancel key 206
The direction key 207 is more intuitive than that of a soft key on the display screen for a beginner user, and can improve the operability of the user in that it can always be used.

FIG. 3 shows a display example of the display-integrated tablet 209. In the figure, 301 is a character input area, and when the user writes a handwritten character in this area with the stylus 202, the handwriting is displayed. Next, when character recognition or conversion is performed based on the handwriting data, the standard font image of the coded characters is displayed in place of the writing data. At 302, “29 Yoshida-cho, Totsuka-ku, Yokohama-shi
The coded character "2" is displayed. Reference numeral 303 denotes a system area, which is an area in which the system displays candidates when the recognition result or the correction result is corrected. Buttons for receiving user's instructions are also displayed in the system area as indicated by 304 and 305.

Next, the control procedure of the system of one embodiment of the present invention will be described. FIG. 9 shows a control flow. First, when the user inputs a handwritten character in the character input area of the display-integrated tablet 209 (step 901), the handwriting is displayed as shown in FIG. 4 and the ink data as the handwriting information and the number of input characters are stored in the storage means. (Step 90)
3). After the user writes a desired number of characters, the conversion button displayed in the system area is designated by the stylus pen (step 903), and the system recognizes the ink data stored in the storage unit by the character recognition unit (step S903). Step 904). The character recognition result is registered in the storage means in the format shown in FIG. 8 in FIG.
01 is the above-mentioned ink data. In 802, the number of recognition candidates obtained as a result of character recognition is stored. 803
The character codes of the recognition candidates and the recognition accuracies indicated by the values of 1 to 0 are stored as a pair, and the recognition candidate numbers stored in 802 are sequentially registered in descending order of accuracy. The number of input characters is stored in 804. In the example of FIG.
It shows that “shi”, “ku”, and “n” are recognized with the accuracy of 0.8, 0.7, and 0.4, respectively, with respect to the input of the first character in FIG. Also, it shows that only one candidate "Ma" is registered as the recognition candidate for the input of the second character, and "book" and "tree" are registered as the recognition candidates for the input of the third character. . Note that, in the table of FIG. 8, for convenience of explanation, the maximum number of recognition candidates is n and the maximum number of input characters is a secondary array, but the number of recognition candidates 802 is used to create a variable length data format. Can be stored efficiently.

In the conventional system, at this stage, the user selects a desired character from the above-mentioned recognition candidates to confirm the character, and then proceeds to the step of the next Kanji conversion. The character input device of the present invention is characterized by performing kana-kanji conversion using the table shown in FIG. 8 as it is and outputting kanji conversion candidates including recognition candidates at the time of correction. When the system shown in FIG. 8 is completed, the system next converts the recognition result into kanji (step 905). Here, the word "kanji conversion" is used in a broad sense of character conversion, and the conversion result may not be kanji. The system creates searchers for all combinations of recognition results and searches the conversion dictionary. Table 1 shows an example of the conversion dictionary.

[0023]

[Table 1]

This dictionary includes kanji in the heading,
It is possible to match the recognition results as words, including the case where all are entered in kanji. As a result of the conversion using the dictionary of Table 1, the conversion result as shown in Table 2 is obtained from the previous input example regarding the dictionary lookup from the first character.

[0025]

[Table 2]

[Table 2] "Kumagi" and "Numamoto"
Indicates that no conversion result was obtained because there are no candidates in the dictionary. Although not described in detail, in the character input device of this embodiment, the conversion is performed over a plurality of phrases by using the known two-phrase longest matching method, the minimum number of clauses method, or the like. That is, “city” + “Maki”, “Shima” + “book”, etc. are internally created as conversion candidates. In this case, the system replaces the handwriting as the conversion result and displays "Shimamoto", which has the smallest number of clauses, the largest total accuracy of recognition of the searchers, and the highest dictionary registration order (step 906).
The appearance of the screen at this time is shown in FIG.

Here, if the displayed result is the character desired by the user, the series of character input processing ends. However, since the character recognition rate and the conversion rate are not generally 100%, the display result is often not the character desired by the user. At this time, as the possibility of error, there are firstly a recognition error and secondly a homophone error. However, as described above, it may be difficult for the user to understand the cause of the error. Even if it is found which is the cause, if the correction method is different, the user must remember two types of correction methods. Therefore,
In the character input device of the present invention, the correction of characters is handled in a unified manner, and any error can be corrected by the same procedure.

When "Shimamoto" is not the desired character, the user designates "Shimamoto" displayed in FIG. Next, when the correction button 505 is designated, the system activates the correction process (step 908). In the correction process,
The conversion result of Table 2 is created again using the recognition candidates of the table of FIG. 8 held in the storage means (step 90).
9). Next, the system displays the conversion result of Table 2 in the system area as shown in FIG. 6 (step 910). In the character input device of the present embodiment, for the convenience of the user, in addition to the conversion candidates, the recognition candidates of the first character to be corrected and the corresponding katakana when the recognition candidates are hiragana are displayed. Has become. In FIG. 6, 601,6
Reference numeral 02 denotes a button for scrolling when the candidates cannot be displayed at one time. In the case of FIG. 6, when 602 is designated, “poem”, “death”, “ward”, “bitter”, “phrase”, “shi”, “ Shi "
"KU", "KU", "N", and "N" are displayed (step 91).
3). The display order of the candidates at this time is according to the order of the total accuracy of the searchers and the order of dictionary registration in this embodiment, but it is also possible to reflect the frequency of use of the user's past words and the like. The character string to be replaced in the character input area is underlined to distinguish it from other character strings.

Here, the user selects a correct candidate from the system area candidates (steps 911, 91).
2) The desired character string can be obtained (step 91).
4). By doing this, for example, the user inputs the first character as "ku" and the desired character string is "Kumamoto".
If it is, all you have to do is specify "Kumamoto", or if the desired character string is "Shimamoto" which is a different notation of "Shimamoto", just specify "Shimamoto" in the same way. It can be modified to the correct string. FIG. 7 shows a screen display example when the user selects “Kumamoto”.

The embodiment of the present invention has been described above. However, if the present small-sized information equipment is to be provided at a low cost,
Since the processing capacity of the PU is low, if the above system is mounted as it is, there may be a problem in practical use. In such a case, part of the present invention may be modified. That is, generally, if the response to the conversion instruction is made quick, the user does not feel any pain for the correction instruction even if it takes some time. For example, in the case of outputting the character string of the first candidate in response to the conversion instruction instead of performing character conversion for all recognition candidates, only the first candidate of the recognition result is character converted and the responsiveness is maintained and corrected. When the instructions are matched, the recognition candidate and the conversion candidate may be simultaneously displayed by the above method. Furthermore, the character conversion may not be performed or may be omitted so that the user can reliably select. Also in this case, the conversion instruction step is not necessary as compared with the conventional method, and the method of the present invention is superior. In addition, even when the recognition step and the conversion step are completely separated, the user selects the recognition candidate as shown in FIG.
803 as one candidate and applying the character conversion means of the present invention in consideration of the possibility of forgetting to select a portion not selected by the user, there is no need to re-input as in the conventional case. , Character input can be done efficiently. Further, each of the above-described methods can be arbitrarily switched by the user, and the most efficient input can be performed depending on the proficiency level of the user or the progress of optimization by the learning function of the system.

[0031]

As described above, in the character input device of the present invention, it is possible to omit the operation step of correcting the recognition candidate and reduce the number of operation steps of the user. Moreover, since it is not necessary to separately learn the correction operation for the recognition candidate and the conversion candidate, it is possible to reduce the burden of character input on the user.

[Brief description of drawings]

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

FIG. 2 is an external view of an embodiment of the present invention.

FIG. 3 is a screen configuration diagram of an embodiment of the present invention.

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

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

FIG. 6 is a diagram showing an example of a screen display according to an embodiment of the present invention.

FIG. 7 is a diagram showing an example of a screen display according to an embodiment of the present invention.

FIG. 8 is a diagram showing a state of a storage unit according to an embodiment of the present invention.

FIG. 9 is a control flow diagram of an embodiment of the present invention.

Front page continuation (72) Inventor ▲ Tomomi Hiromi Tadashi, 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa, Ltd.Hitachi, Ltd. Microelectronics Device Development Laboratory (72) Inventor Shigeki Hataya 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa House number Hitachi, Ltd. Microelectronics Device Development Laboratory (72) Inventor Akira Nakajima House number 292 Yoshida-cho, Totsuka-ku, Yokohama City, Kanagawa Prefecture Hitachi Ltd. Microelectronics Device Development Laboratory

Claims (5)

[Claims] 1. A coordinate input means for inputting coordinates, a character recognition means for converting an image or stroke data of a character input from the coordinate input means into a character code, and a character string as a recognition result of the character recognition means. A character conversion means for converting into a character string of 2; a correction instruction means for instructing correction of the converted character; and a display means for displaying the coordinate input result, the recognition result, the conversion result, the recognition candidate, and the conversion candidate. A character input device characterized in that a recognition candidate of a character instructed by the correction instructing means and conversion candidates for all combinations of the recognition candidates are simultaneously displayed and selectable. 2. A conversion instruction means for converting the image or stroke data of the character into a character code, and when the conversion instruction is given, the target image or stroke data is character-recognized and then the recognition is performed. The character input device according to claim 1, wherein character conversion is continuously performed for all combinations of the resulting recognition candidates to determine and display the first candidate. 3. A conversion instruction means for converting the image or stroke data of the character into a character code, and when the conversion instruction is given, character recognition is performed on the target image or stroke data and then the recognition. Character conversion is performed for the first candidate among the resulting recognition candidates,
The character input device according to claim 1, wherein the first candidate is determined and displayed. 4. A conversion instruction means for converting the image or stroke data of the character into a character code, and when the conversion instruction is given, character recognition is performed on the target image or stroke data and then the recognition. The first candidate is displayed as it is without performing character conversion for the first candidate among the resulting recognition candidates.
Character input device described. 5. A step of inputting a handwritten character, a step of instructing recognition of the handwritten character and conversion into kanji, and a step of instructing correction of the character when the displayed result is different from the desired one. A character input method, comprising: selecting a desired character from the homophones and the recognition candidates that are simultaneously displayed as correction candidates for the correction-instructed character, and a step of selecting a desired character.