JPS5827549B2 - handwritten character recognition device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention provides a handwritten character recognition device, particularly a handwritten character recognition device equipped with a tablet handwriting input section, which includes a recognized character correction function for correcting recognition results.
The present invention relates to a handwritten character recognition device that can display, for example, a menu using a recognition dictionary used for handwritten character recognition based on information from an input unit of the recognized character correction function.

Character recognition devices have a correction mechanism that corrects unreadable characters and inputs correct characters.

In recognition devices that read characters with a small number of characters, such as alphabets, numbers, and kana, a method is adopted in which corrections are made from the keyboard.

However, when dealing with kanji, since the number of characters is large and inputting directly on a keyboard is inconvenient, a kana keyboard is generally used to read the kanji (tuning fork is kun).
The Kana-Kanji conversion method is adopted, in which the user inputs a homonym and selects it by displaying a menu of homophones on the display.

For this purpose, we have a kana/kanji conversion dictionary.

On the other hand, in a handwritten character recognition device using a tablet, characters are handwritten on the surface of a digitizer called a tablet, and the characters are recognized using the movement of the writing point and the signal of the vertical movement of the pen.

That is, the number of strokes (number of strokes) is determined from the vertical movement of the pen, and the order of strokes is detected from the movement of the pen point, and the stroke order of each character is recognized.

For this reason, in the above-mentioned handwritten character recognition device,
It has a recognition dictionary classified according to the number of strokes.

The present invention makes it possible to perform correction processing by using the recognition dictionary when the unreadable characters are corrected,
The purpose is to unify the kanji dictionaries provided in handwritten character recognition devices.

To this end, the handwritten character recognition device of the present invention is equipped with a tablet handwritten input unit, and compares characters including kanji characters handwritten from the input unit with a recognition dictionary and outputs the results as recognition result information. , a recognized character correction function including an input section, a display, and a correction processing section is added, and a display control section in the correction processing section corrects characters based on at least the number of strokes based on display instruction information instructed by the input section. is configured to index the recognition dictionary in which are arranged, display the characters of the index result on the display, and output the designated character based on the display result as modified result information in place of the recognition result information. It is characterized by being configured as follows.

This will be explained below with reference to the drawings.

Fig. 1 shows the configuration of an embodiment of a conventional handwritten character recognition device with a correction function using a menu display, Fig. 2 shows an example of a kana/kanji conversion dictionary, Fig. 3 shows the configuration of an embodiment of the present invention, and Fig. 4 3 shows an embodiment of the recognition dictionary used in the configuration shown in FIG. 3, and FIG. 5 shows an embodiment of the recognition dictionary used when a menu display using pronunciation is also used.

In FIG. 1, 1 is a recognition processing section, 2 is a tablet input section, 3 is a correction processing section, 4 is a keyboard, 5 is a display, 6 is a recognition section, 7 is a stroke number detection section, 8 is a feature extraction section, 9
10 is a recognition dictionary, 10 is a collation unit, 11 is a display control unit, 12 is a kana/kanji conversion dictionary, 13 is a character generator, and 14 is an OR logic unit.

The number of strokes of handwritten characters input from the tablet input unit 2 is detected by the stroke count detection unit 7, and features such as stroke order are extracted by the feature extraction unit.

Then, based on these results, for example, first narrow down the candidates from the characters in the recognition dictionary based on the stroke count information, and then narrow down the candidates based on the above feature information, such as ``characters with many strokes and in what stroke order.'' Finally, the matching unit 10 performs recognition processing in correspondence with more specific feature information, and the result is presented as a "recognition result" as an answer.

If an unreadable character occurs at this time, the recognition section 6 notifies the display control section IH of the correction processing section 3.

The occurrence of this unreadable character is displayed on the display 5, and the operator inputs the pronunciation of the character in kana in order to correct the unreadable character.

The display control unit 11 displays the input pronunciation (corresponding to kana/
The Kanji conversion dictionary 12 is accessed, and if homophones exist, each of these homophones is displayed as a menu on the display 5.

That is, the kana/kanji conversion dictionary 12 is arranged in accordance with the readings of kanji, as shown in FIG. 2, for example, and is stored in the form of codes representing each kanji.

Now, when the operator inputs the kana ``A'' from the keyboard 4 to correct the kanji ``pressure'', the kanji ``love'', ``evil'', ``pressure'', etc. shown in Figure 2 will be changed into characters. The characters are converted into model numbers via the generator 13 and displayed as a menu on the display 5.

When the operator looks at this menu display and specifies the kanji character ``pressure'' using the keyboard 4, the code corresponding to the kanji character ``pressure'' passes through the OR logic section 14 and is used as an answer in the form of correcting the unreadable characters. Output.

However, in the case of this method, the recognition dictionary 9, which is arranged with the number of strokes as headings, and the kana/kanji conversion dictionary 12, which is arranged with readings as headings, are placed side by side.

FIG. 3 shows the configuration of an embodiment of the present invention, and the reference numerals in the figure correspond to those in FIG. 1, so to speak.
This is an abbreviated version of the Kanji Conversion Dictionary 12.

In this case, in response to the occurrence of unreadable characters, the kanji ``pressure''
In order to correct this, the operator is made to input from the keyboard 4 the number of strokes of the kanji character "pressure", "5".

As a result, the display control unit 11 indexes the recognition dictionary 9 based on the number of strokes "5" and displays a menu of kanji with the same number of strokes on the display 5.

That is, the recognition dictionary 9 is arranged in accordance with the number of strokes, as shown in FIG. 4, for example, and stores codes representing each kanji character as well as characteristics necessary for recognition processing.

The operator is now using the keyboard 4 to correct the kanji ``pressure''.
If the number of strokes "5" is input from , the display control unit 11 indexes the recognition dictionary 9 using the number of strokes "5".

That is, the recognition dictionary 9 is accessed in the same way as when the number of strokes is given from the tablet input section 2.

And from the recognition dictionary 9, the kanji "world" with the number of strokes "5",
Codes such as "main", "other", "pressure", etc. are read out and transferred to the display control section 11.

The display control unit 11 uses these codes to display characters and characters.
A display character pattern is extracted from the generator 13 and a menu is displayed on the display 5.

When the operator looks at the menu display and specifies the Chinese character ``pressure'' using the keyboard 4, a code corresponding to the Chinese character ``pressure'' is sent through the OR logic section 14 and output as an answer in the form of correcting the unreadable characters.

As the recognition dictionary 9, kanji can be arranged in the form of headings in which the number of strokes and the reading are grouped together, as shown in the configuration shown in FIG.

Needless to say, in this case as well, the code corresponding to each kanji and its characteristics are stored.

When the configuration shown in FIG.
It is possible to specify the number of strokes as ``5'' and the reading ``A'' corresponding to ``pressure''.

Even in this case, the recognition dictionary 9 is still indexed according to instructions from the display control unit 11.

However, when displaying the menu, the number of strokes is read as "5".
The number of characters to be selected is limited to only kanji with "A", and the time required for the correction process is greatly reduced.

As explained above, according to the present invention, the display control unit 11
Since the recognition dictionary 9 existing in the recognition unit 6 can be indexed, it becomes possible to unify the dictionaries created in the handwritten character recognition device.

Furthermore, when performing the correction process, the operator inputs the number of strokes of a character, and the operator learns the correct number of strokes of a character to be input from the tablet input unit 2 for the original recognition process.

[Brief explanation of the drawing]

Fig. 1 shows the configuration of an embodiment of a conventional handwritten character recognition device with a correction function using a menu display, Fig. 2 shows an example of a kana/kanji conversion dictionary, Fig. 3 shows the configuration of an embodiment of the present invention, and Fig. 4 is an example configuration of a recognition dictionary used in the configuration shown in FIG.
FIG. 5 shows an embodiment of the configuration of a recognition dictionary used when menu display using pronunciations is also used. In the figure, 1 is the recognition processing unit, 2 is the tablet human power unit, 3 is the correction processing unit, 4 is the keyboard, 5 is the display, 6 is the recognition unit, 7 is the number of strokes detection unit, 8 is the feature extraction unit, and 9 is the recognition dictionary , 10 represents a matching section, 11 a display control section, 13 a character generator, and 14 an OR logic section.

Claims (1)

[Claims] 1. A handwritten character recognition device that is equipped with a tablet handwriting human power section and that compares characters including kanji characters handwritten from the input section with a recognition dictionary and outputs them as recognition result information, including an input section, a display, and a correction processing section. In addition to adding a recognition character correction function, the display control unit in the correction processing unit indexes the recognition dictionary in which characters are arranged based on at least the number of strokes, based on display instruction information instructed by the input unit. Handwritten character recognition characterized in that the characters of the index result are displayed on the display and the characters instructed based on the display result are output as corrected result information in place of the recognition result information. Device.