JPS6316796B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an improvement in a recognition mode designation method in a real-time character recognition device.

Since Japanese sentences use a wide variety of characters such as kanji, hiragana, katakana, and alphanumeric characters, it is extremely difficult to recognize a mixture of these characters. For example, the kanji ``工'' and the katakana ``e'' cannot be distinguished. Therefore, in order to recognize these characters without error, it is necessary to inform the recognition unit in advance of the type of character to be input using a function key or the like. In the present invention, teaching the type of character to be input is called recognition mode. However, since changing the recognition mode is a very unnatural action for some Iruma users, they often forget this action and input a different type of character to achieve recognition. Such input errors usually cannot be recovered from until the operator notices the input error.

Therefore, the present invention does not specify the recognition mode while inputting characters, but starts recognition by specifying the recognition mode after inputting at least one continuation sentence. The purpose is to prevent misrecognition of characters.

For this reason, the recognition mode designation method in the real-time sentence recognition device of the present invention requires a tablet for handwriting characters to be input, and a predetermined recognition mode among the recognition modes such as Kanji hiragana mode, katakana mode, and alphanumeric mode. At least an input means for specifying a recognition mode, a dictionary corresponding to the recognition mode such as a kanji hiragana dictionary, a katakana dictionary, and an alphanumeric dictionary, and a dictionary corresponding to the mode is specified by the input means. The real-time handwritten character recognition device recognizes characters handwritten on the tablet with reference to After a sentence unit is input, a position corresponding to each of the recognition modes in the sentence is indicated on the tablet, and a recognition mode corresponding to the position is input from the input means. It is something to do. The present invention will be explained below with reference to the drawings.

FIG. 1 is a diagram showing an example of a tablet input surface used in the present invention, FIG. 2 is a diagram showing an actual example of handwritten character input, and FIG. 3 is a block diagram of one embodiment of the present invention. FIG. 4 is a diagram showing an example of automatic recognition mode change.

In the figure, 1 is a tablet, 2 is manuscript paper,
3 indicates a recognition mode change key, and 4 indicates a stylus pen.

As shown in FIG. 1, when inputting characters, a manuscript paper 2 is placed on the input surface of a tablet 1, and Japanese characters are input using an attached stylus pen 4. At this time, coordinate point series information corresponding to the movement of the stylus pen and pen up/down information are sent to the recognition device.

Figure 2 shows an example of actual character input. For example, when inputting the words ``We are faced with the need to input Japanese into a computer,'' the operator would It is necessary to perform certain operations.

(i) At point P ₁ , press the kanji kana part of recognition mode change key 3.

(ii) Fill in "We Japanese".

(iii) At point _P2 , press the Katakakana part of the recognition mode change key 3.

(iv) Write “computer”.

(v) Press the Kanji/Kana part of recognition mode change key 3.

(vi) Write ``I am faced with the need to input Japanese.''

Characters are input into the recognition device as described above, but in actual character input operations, it is common to forget to specify the recognition mode at point P ₂ or point P _{3 .} If you forget to specify this part, the "computer" part will be recognized in kanji hiragana mode, and if you forget to specify it in point _P3 , the following will be recognized in katakana mode, and it will not be recognized at all. I can't do it, and I end up misrecognizing it. The present invention has been made to prevent such misrecognition.

FIG. 3 is a block diagram of one embodiment of the present invention. In the figure, the same symbols as in FIG. 1 indicate the same things, and 5 is an input control circuit, 6 is a feature extraction circuit, 7 is a matching circuit, 8 is a mode change control circuit, 9 is a position detection circuit, and 10 is a matching circuit. is a stroke number detection circuit, 11 is an input buffer, 12 is an output buffer,
Reference numeral 13 indicates a temporary storage circuit, and reference numeral 14 indicates a dictionary.

The input control circuit 5 performs cutting out of characters, extraction of representative points, detection of recognition mode, and the like.

The feature extraction circuit 6 is, for example, a relative position matrix creation circuit that creates a relative position matrix between representative points.

The matching circuit 7 compares the characteristics of the character type in the dictionary with the characteristics created by the feature extraction circuit 6 to determine the type of input character.

The dictionary 14 is divided into a kanji hiragana group, a katakana group, an alphanumeric group, etc., and is further divided into subgroups according to the number of strokes.

When the recognition mode information and stroke count information are input, the characteristics of all the corresponding character types are sent to the matching circuit 7 and compared with the characteristics of the input character.

The characters written on the tablet 1 are once stored in the temporary storage circuit 13 via the input buffer 11.

The position detection circuit 9 transfers the coordinate values corresponding to the character cutout obtained by the input control circuit 5 to the temporary storage circuit 1.
Input it in addition to the character data input in step 3. The recognition mode detection circuit flags a portion of the temporarily stored coordinate values of the input character that corresponds to the most advanced coordinate value of the input character that requires a recognition mode change.

Reference numeral 12 denotes an output buffer, which inputs the read characters to the feature extraction circuit 6 and changes the recognition mode of the dictionary 14 while referring to the flag indicating the recognition mode.

Next, the recognition mode designation method of the present invention will be explained.

First, a sentence as shown in FIG. 2 is inputted from the tablet 1 in its entirety, and is sequentially stored in the temporary storage circuit.

Thereafter, the position corresponding to each recognition mode is input as a point on the tablet, and the recognition mode corresponding to that point is input using the recognition mode change key 3 shown in FIG.

In this manner, recognition is started by pressing a recognition start key (not shown).

That is, the temporarily accumulated information is read out from 13 and input to the feature extraction circuit 6, and a recognition mode signal is given to the dictionary 14. In this way, the matching circuit 7 performs matching and outputs the answer.

In this manner, the present invention allows the recognition mode to be specified later, thereby eliminating designation errors and improving the recognition rate.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing an example of a tablet input surface used in the present invention, Fig. 2 is a diagram showing an actual example of handwriting input, and Fig. 3 is a block diagram of one embodiment of the present invention. . 1 is a tablet, 2 is a manuscript paper, 3 is a recognition mode change key, 4 is a stylus pen, 5 is an input control circuit, 6 is a feature extraction circuit, 7 is a matching circuit, 8 is a mode change control circuit, 9 is a position detection circuit, 10 is a stroke number detection circuit, 11 is an input buffer,
12 is an output buffer, 13 is a temporary storage memory circuit,
14 each indicate a dictionary.

Claims (1)

[Scope of Claims] 1. A tablet for handwriting characters to be input; an input means for specifying a predetermined recognition mode among recognition modes such as a kanji hiragana mode, a katakana mode, and an alphanumeric mode; and a kanji hiragana dictionary. a katakana dictionary, and a dictionary corresponding to a recognition mode such as an alphanumeric dictionary, and specifying a mode with the input means and referring to the dictionary corresponding to the mode to read characters handwritten on the tablet. The real-time handwritten character recognition device has means for temporarily storing characters handwritten on the tablet, and after at least one sentence unit of characters to be input is input, Recognition mode designation in a real-time handwritten character recognition device, characterized in that a position corresponding to each of the recognition modes is indicated on the tablet, and a recognition mode corresponding to the position is input from the input means. Method.