JPH08106513A - Hand-written character recognition device - Google Patents

Abstract

PURPOSE: To easily and quickly register characterized hand-written characters in a dictionary to improve the character recognition rate by performing the character recognition processing of characterized hand-written characters without preliminarily registering them in the dictionary and registering them in the dictionary in the case of erroneous recognition. CONSTITUTION: A dictionary 5 where candidate characters are registered correspondingly to the input pattern of a hand-written character, a character recognition means 3 which collates the input pattern of the hand-written character with input patterns in the dictionary 5 to display the candidate characters of character patterns having high similarities, and a register means 4 which allows the input of the hand-written character again to display the candidate characters of input patterns having high similarities from the dictionary 5 in the case of a failure to find a desired character in candidate characters obtained by collation and additionally registers a correct character to initial candidate characters in the dictionary 5 at the time of selecting the correct character from displayed candidate characters are provided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a handwritten character recognition device.

[0002]

2. Description of the Related Art Conventionally, in a handwritten character recognition process, a handwritten character is recognized by using a fixed dictionary that is provided as a standard for character recognition. Recently, in order to correctly recognize handwritten characters with habits, a new word dictionary is provided in addition to a fixed dictionary, and handwritten characters with habits are registered in advance in the new word dictionary, and this new word dictionary is also used for recognition. I try to increase the rate.

[0003]

As described above, when a new word dictionary is provided for recognizing habitual handwritten characters, as an operation for registering habitual handwritten characters in this new word dictionary, "Register button Tap, select character type,
It is necessary to individually perform "character selection", "handwritten character input", and "tap registration button", and especially when it comes to registering many handwritten characters with many habits, extremely troublesome operations are repeated. In addition to the problem that it becomes, there is a possibility that the habit of typing without anything comes out and the registration becomes meaningless. For this reason, there is a problem that it is faster to re-input the stroke order and the like with care, rather than registering handwritten characters with habit in the new word dictionary beforehand.

The present invention solves these problems.
Without registering handwritten characters with habits in the dictionary in advance, register them when you misrecognize them by incorporating them in the character recognition process, and easily and quickly register the handwritten characters with habits for character recognition. The purpose is to improve the rate.

[0005]

FIG. 1 shows a block diagram of the principle of the present invention. In FIG. 1, the input device 1 is for inputting handwritten characters and candidate characters.

The character recognition device 2 is for recognizing handwritten characters, and is composed of a character recognition means 3, a registration means 4, etc. here. Character recognition means 3
Is for collating the input pattern of the input handwritten character with the input pattern in the dictionary 5 and extracting candidate characters of the input pattern having a high degree of similarity.

The registration means 4 is for additionally registering new candidate characters in the dictionary 5. The dictionary 5 registers candidate characters in association with input patterns. The display device 6 displays an input pattern of handwritten characters, displays candidate characters, and the like.

[0008]

According to the present invention, as shown in FIG.
Collates the input pattern of the characters input by handwriting with the input pattern of the dictionary 5, extracts the candidate characters of the character pattern having a high degree of similarity and displays them on the display device 6, and when the desired character is not found, the handwriting is input again. In response to the fact that the candidate character of the matching input pattern is displayed from the dictionary 5 and the correct character is selected, the registration means 4 additionally registers this correct character in the original candidate character in the dictionary 5. I have to.

Further, the character recognition means 3 collates the input pattern of the handwritten character with the input pattern of the dictionary 5, displays the candidate characters of the input pattern having a high degree of similarity, and finds the desired character in this candidate character. If the correct character is input from the input device when there is not, the registration means 4
Is additionally registered as the initial candidate character in the dictionary 5.

Therefore, without registering handwritten characters with habits in the dictionary 5 in advance, additional registration is performed in the dictionary 5 when misrecognized by incorporating during the character recognition process, and handwritten characters with habits are registered. You can register your dictionary easily and quickly,
It is possible to improve the character recognition rate.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the construction and operation of an embodiment of the present invention will be described in detail with reference to FIGS.

FIG. 1 shows a block diagram of an embodiment of the present invention.
In FIG. 1, the dictionary 5 is associated with a character pattern in a standard stroke order of handwritten characters and registers candidates for character recognition at that time. For example, in the case of pattern 1, candidate characters are registered as shown below in association with the character pattern of the stroke order when handwriting the character "ra".

.., J, B ... Similarly, the candidate characters are registered as shown below in association with the character pattern in the stroke order when the character "5" is handwritten.

.5, 5, ..., etc. Here, "ra" is newly registered in the present invention (described later with reference to the flowchart of FIG. 2).

Next, the operation of the configuration of FIG. 1 will be described in detail according to the order shown in the flowchart of FIG. 2 and with reference to FIG. FIG. 2 shows a flowchart for explaining the operation of the present invention.

In FIG. 2, S1 waits for character input. S2 is displayed in the character input frame. This corresponds to the fact that the handwritten character is written on the tablet which is the input device, and the writing pressure is detected to detect the input pattern of the writing order of the handwritten character. The handwritten character "ra" is displayed in the character input box of a).

In step S3, character recognition is performed by referring to the dictionary. This is to perform pattern matching between the input pattern of the handwritten character displayed in the character input frame in S2 and the input pattern in the dictionary 5, and retrieve the candidate character registered in association with the input pattern having the highest degree of similarity. .

In step S4, the recognized character is displayed in the recognized character frame and the character in the character input frame is saved. This is to display the first character of the candidate character extracted in S3 in the recognized character frame as, for example, "ra" shown in FIG. 3B, and save the character pattern of the handwritten character in the character input frame. .

In step S5, it is determined whether a new character has been input. If YES, the next input is made, so S2
Return to and repeat. On the other hand, if NO, the process proceeds to S6.
In S6, it is determined whether the recognized character is a tap. This determines whether the operator has tapped the recognition character displayed in the recognition character frame in S4. If YES, the next candidate character is displayed in S7, that is, the next candidate character, for example, "5, 5, Ta, _Ta ..." Is displayed in the next candidate character frame as shown in (c) of FIG. Then, the process proceeds to S8. On the other hand, if NO, the process returns to S5.

In step S8, it is determined whether a candidate character has been designated. This determines whether any candidate character is designated from the next candidate characters displayed in the next candidate character frame. Yes
In this case, since the next candidate character has a correct character, the character display selected in the recognition character frame in S14 and the candidate character are deleted. On the other hand, in the case of NO, it is not specified even if the correct candidate character is not displayed, so the process proceeds to S9.

In step S9, it is determined whether registration has been tapped.
This determines whether the "registration button" displayed on the screen has been tapped. If yes, then continue with S10. NO
In the case of, it returns to S8 and repeats.

In step S10, the handwritten pattern and the next input frame are displayed. This is YES in S9, and since registration is tapped because there is no correct character among the next candidate characters, FIG.
As shown in (d) of, the saved handwriting pattern is displayed and the next input frame is displayed.

In S11, a series of input processing is performed. This is because the handwritten characters are input again in the correct stroke order, the input pattern is collated with the input pattern of the dictionary 5,
The candidate character with the highest degree of similarity is displayed, or the correct character is keyed in from the keyboard which is the input device 1 and displayed.

In S12, tap registration. This is S
With the correct characters displayed in 11, tap Register,
Instruct to register. In S13, it is registered in the dictionary of the corresponding input character. This is the correct character registered in S12,
It is additionally registered in the candidate characters in the dictionary 5.

As described above, the handwritten character is input, the input pattern is displayed, the input pattern is collated with the input pattern in the dictionary 5, the candidate character of the input pattern having the highest degree of similarity is extracted and used as the character recognition result. If there is no correct character among these candidate characters, carefully input the handwritten characters again in the correct stroke order, check the input pattern with the input pattern in the dictionary 5, and display the candidate character with the highest similarity. Then, when the registration is tapped with the correct characters being input, or with the correct characters being input from the keyboard or being displayed by entering the Roman-Kana conversion, additional registration is added to the initial candidate characters in the dictionary 5. .

As described above, the input pattern of the handwritten character is collated with the input pattern in the dictionary 5, the candidate character with the highest degree of similarity is displayed, and when there is no correct character in this candidate character, again, Input the handwritten characters in the correct stroke order, check the input pattern in the dictionary 5 and display the candidate character with the highest degree of similarity, specify the correct one from them, or enter the correct character from the keyboard or Roman. Corresponding to the input by conversion, it is additionally registered as a candidate character in the dictionary 5. As a result, the correct character is associated with the input pattern of the handwritten character with a habit and the correct character 5
It becomes possible to additionally register in it, and even if it is a handwritten character with a habit from the next time, it will be displayed in the candidate character,
It is possible to easily select and input the character recognition result.

FIG. 3 shows an operation explanatory diagram of the present invention. This shows how it is displayed on the screen of the display device 6. FIG. 3A shows a state in which the handwritten character “ra” is input and displayed in the character input frame. Here, each frame is
As shown in the figure from top to bottom: ・ Recognized character frame ・ Next candidate character frame ・ Character input frame. Here, the recognition character frame is a frame in which the character pattern of the handwritten character is collated with the input pattern in the dictionary 5 and the character having the highest priority among the candidate characters of the input pattern having the highest similarity is displayed as the recognition character. Is. The next candidate character frame is a frame for displaying a candidate character string. The character input frame is a frame for displaying an input pattern of handwritten characters.

FIG. 3B shows a state in which the character "5", which is the result of character recognition when the handwritten character "ra" in FIG. 3A, is displayed in the recognized character frame. In this case, since it is a handwritten character with a habit, it is erroneously recognized.

FIG. 3C shows a state in which the next candidate is displayed in the recognized character frame in the state of FIG. 3B and all the displayable candidate characters are displayed in the next candidate character frame. FIG. 3D shows a state in which registration is tapped. This is shown in Figure 2.
The state at the time of S10 is displayed: -The first character of the candidate characters is displayed in the recognized character frame-All the candidate characters are displayed in the next candidate character frame-The character input frame is displayed Displays the character pattern when the first handwritten character in FIG. 3A is input. Then, one more character input frame is displayed on the right side.

FIG. 3E shows that the handwritten character "R" is input again in the correct stroke order in the state of FIG. 3D, and the character pattern at that time is displayed in the character input box on the right side. ,
The state in which the character “R” recognized at that time is displayed in the recognized character frame is shown. In this case, since the handwritten character "ra" is input in the correct stroke order, the correct recognition result "ra" is displayed in the recognition character frame. Incidentally, in the state of FIG. 3 (e), the correct character "ra" may be keyed in by the keyboard in the recognition character frame, or the correct character "ra" may be converted by Roman-kana conversion of pen input. You may input it.

FIG. 3 (f) shows the initial ((a) of FIG. 3) character pattern displayed in the left character input frame by tapping the "register button" in the state of (e) of FIG. A state in which the correct character "ra" displayed in the recognized character frame is additionally registered as a candidate character in the dictionary 5 is shown.

3 (g), after additional registration in FIG. 3 (f), similarly to FIG. 3 (a), additionally registered in the next candidate character frame when a quirky handwritten character is input. Was done
Is displayed. As a result, it is possible to select the correct "ra" from the next candidate character frame, display it in the recognized character frame, and input it.

[0033]

As described above, according to the present invention,
Since a handwritten character having a habit or the like is not registered in the dictionary 5 in advance, it is incorporated in the character recognition process, and when it is erroneously recognized, it is additionally registered in the dictionary 5,
It is possible to easily and quickly register a dictionary of handwritten characters with habits and improve the character recognition rate.

[Brief description of drawings]

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

FIG. 2 is a flowchart explaining the operation of the present invention.

FIG. 3 is an operation explanatory diagram of the present invention.

[Explanation of symbols]

 1: Input device 2: Character recognition device 3: Character recognition means 4: Registration means 5: Dictionary 6: Display device

Claims (3)

[Claims] 1. A dictionary (5) for registering candidate characters in association with an input pattern of handwritten characters, an input pattern of handwritten characters, and the dictionary (5).
And a character recognition means (3) for displaying a candidate character of a character pattern having a high degree of similarity by collating with the input pattern, and handwriting input again when a desired character is not found in the candidate character and inputting from the dictionary (5). A registration unit (4) for additionally registering the correct character to the original candidate character in the dictionary (5) in response to the fact that the candidate character of the input pattern having a high degree of similarity is displayed and the correct character is selected. And a handwritten character recognition device. 2. A dictionary (5) in which candidate characters are registered in advance in association with handwritten character input patterns, an input pattern of handwritten characters, and the dictionary (5).
A character recognition means (3) for displaying a candidate character of a character pattern having a high degree of similarity by collating it with the input pattern of (3) and a correct character being input from the input device when a desired character is not found in the candidate character. Correspondingly, a handwriting character recognition device comprising: a registration means (4) for additionally registering to an initial candidate character in the dictionary (5). 3. The handwritten character recognition device according to claim 1 or 2, wherein the input pattern erroneously recognized during the text input process is used as it is for the registration process.